Skip to content

Listings

All listings: NanoGPT

Every listing for the models on page 3 of 40 of NanoGPT, including each model's other versions and routes. Back to NanoGPT

Sold by NanoGPT

249 listings on this page sold by NanoGPT.

Select a score for details.

About these columns
llminfo IQ
llminfo IQ (capability index). How llminfo IQ is computed
Coding IQ
llminfo Coding IQ (coding capability, its own scale). How Coding IQ is computed
Value
Value for money. 50 means as expected for its results. Above 50, cheaper per task. How Value for money works
Yap
Tokens used per coding task vs. models with the same results. Includes agents that write code. Chattiness on maths doesn't predict it. How the Yap score works
Modelllminfo IQCoding IQInput $/1MOutput $/1MMax Input TokensValueYapEpoch Capabilities Index
Gemini 3.5 Flash
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGemini 3.8 FlashWe compared release dates and version names in our catalogue.
135.9
From 393 benchmarks. Plenty of data behind it.
113 ±5
Coding IQ 113.3, 95% interval 109–118. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.50cache read $0.15$9.001M
50.2
Fair value (50.2). Costs per task about what its benchmark results predict. Plenty of data behind it.
77 ±6
Chatty (77 ±6). Generates more tokens than models with similar coding results. Measured at default effort. A fair amount of data behind it.
155.3
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Gemini 3.5 Flash
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGemini 3.8 FlashWe compared release dates and version names in our catalogue.
google/gemini-3.5-flash-thinking · thinking
135.9
From 393 benchmarks. Plenty of data behind it.
113 ±5
Coding IQ 113.3, 95% interval 109–118. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.50cache read $0.15$9.001M
50.2
Fair value (50.2). Costs per task about what its benchmark results predict. Plenty of data behind it.
77 ±6
Chatty (77 ±6). Generates more tokens than models with similar coding results. Measured at default effort. A fair amount of data behind it.
155.3
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Qwen3.8 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
135.9
From 44 benchmarks. Plenty of data behind it.
Not enough data yet$0.15cache read $0.016$0.47cache write $0.201M
65.8
Good value (65.8). Cheaper per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate.
Not measuredNot measured
Qwen3.8 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
🇸🇬 SG
Inference by Alibaba Cloud
135.9
From 44 benchmarks. Plenty of data behind it.
Not enough data yet$0.1575cache read $0.0168$0.4935Not published by the provider
65.8
Good value (65.8). Cheaper per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate.
Not measuredNot measured
Qwen3.8 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
bf16 · GLB
Inference by Uomi
135.9
From 44 benchmarks. Plenty of data behind it.
Not enough data yet$0.1654$0.5182Not published by the provider
65.8
Good value (65.8). Cheaper per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate.
Not measuredNot measured
GPT-5.6 Terra
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGPT-6.1 TerraWe compared release dates and version names in our catalogue.
135.7
From 329 benchmarks. Plenty of data behind it.
121 ±5
Coding IQ 120.9, 95% interval 116–126. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$2.00cache read $0.20$12.00cache write $2.501.1M
60.4
Good value (60.4). Cheaper per task than its benchmark results predict. Plenty of data behind it.
48 ±10
Typical (48 ±10). Generates about as many tokens as models with similar coding results. Measured at max effort. A fair amount of data behind it.
160.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GPT-5.6 Terra
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGPT-6.1 TerraWe compared release dates and version names in our catalogue.
openai/gpt-5.6-terra-pro · pro
135.7
From 329 benchmarks. Plenty of data behind it.
121 ±5
Coding IQ 120.9, 95% interval 116–126. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$2.00cache read $0.20$12.00cache write $2.501.1M
60.4
Good value (60.4). Cheaper per task than its benchmark results predict. Plenty of data behind it.
48 ±10
Typical (48 ±10). Generates about as many tokens as models with similar coding results. Measured at max effort. A fair amount of data behind it.
160.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GPT-6 Luna
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
135.6
From 227 benchmarks. Plenty of data behind it.
117 ±6
Coding IQ 116.9, 95% interval 111–123. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.10cache read $0.01$0.50cache write $0.1251.1M
63.4
Good value (63.4). Cheaper per task than its benchmark results predict. Plenty of data behind it.
51 ±11
Typical (51 ±11). Generates about as many tokens as models with similar coding results. Measured at max effort. Little data behind it, so treat it as a rough estimate.
156.5
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GPT-6 Luna
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
openai/gpt-6-luna-pro · pro
135.6
From 227 benchmarks. Plenty of data behind it.
117 ±6
Coding IQ 116.9, 95% interval 111–123. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.10cache read $0.01$0.50cache write $0.1251.1M
63.4
Good value (63.4). Cheaper per task than its benchmark results predict. Plenty of data behind it.
51 ±11
Typical (51 ±11). Generates about as many tokens as models with similar coding results. Measured at max effort. Little data behind it, so treat it as a rough estimate.
156.5
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GPT-5.6 Luna
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGPT-6 LunaWe compared release dates and version names in our catalogue.
135.2
From 344 benchmarks. Plenty of data behind it.
112 ±10
Coding IQ 112.4, 95% interval 102–123. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.20cache read $0.02$1.20cache write $0.251.1M
66.2
Good value (66.2). Cheaper per task than its benchmark results predict. Plenty of data behind it.
52 ±8
Typical (52 ±8). Generates about as many tokens as models with similar coding results. Measured at max effort. A fair amount of data behind it.
156.7
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GPT-5.6 Luna
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGPT-6 LunaWe compared release dates and version names in our catalogue.
openai/gpt-5.6-luna-pro · pro
135.2
From 344 benchmarks. Plenty of data behind it.
112 ±10
Coding IQ 112.4, 95% interval 102–123. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.20cache read $0.02$1.20cache write $0.251.1M
66.2
Good value (66.2). Cheaper per task than its benchmark results predict. Plenty of data behind it.
52 ±8
Typical (52 ±8). Generates about as many tokens as models with similar coding results. Measured at max effort. A fair amount of data behind it.
156.7
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Ling 3.1 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionLing 3.2 FlashWe compared the version names in our catalogue.
135.2
From 16 benchmarks. A fair amount of data behind it.
Not enough data yet$0.075cache read $0.015$0.22262K
69.6
Good value (69.6). Cheaper per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate.
Not measuredNot measured
Ling 3.1 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionLing 3.2 FlashWe compared the version names in our catalogue.
fp4 · 🇺🇸 US
Inference by DeepInfra
135.2
From 16 benchmarks. A fair amount of data behind it.
Not enough data yet$0.315cache read $0.063$0.945Not published by the provider
69.6
Good value (69.6). Cheaper per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate.
Not measuredNot measured
Qwen3.7 Max
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionQwen4 MaxWe compared release dates and version names in our catalogue.
134.9
From 225 benchmarks. Plenty of data behind it.
112 ±5
Coding IQ 112.2, 95% interval 107–117. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.475cache read $0.295$4.425cache write $1.84381M
44.2
Poor value (44.2). Costs more per task than its benchmark results predict. Plenty of data behind it.
55 ±9
Typical (55 ±9). Generates about as many tokens as models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Qwen3.7 Max
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionQwen4 MaxWe compared release dates and version names in our catalogue.
qwen/qwen3.7-max:thinking · thinking
134.9
From 225 benchmarks. Plenty of data behind it.
112 ±5
Coding IQ 112.2, 95% interval 107–117. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.475cache read $0.295$4.425cache write $1.84381M
44.2
Poor value (44.2). Costs more per task than its benchmark results predict. Plenty of data behind it.
55 ±9
Typical (55 ±9). Generates about as many tokens as models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Qwen3.7 Max
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionQwen4 MaxWe compared release dates and version names in our catalogue.
🇸🇬 SG
Inference by Alibaba Cloud
134.9
From 225 benchmarks. Plenty of data behind it.
112 ±5
Coding IQ 112.2, 95% interval 107–117. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$2.625cache read $0.525$7.875Not published by the provider
44.2
Poor value (44.2). Costs more per task than its benchmark results predict. Plenty of data behind it.
55 ±9
Typical (55 ±9). Generates about as many tokens as models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Qwen3.7 Max
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionQwen4 MaxWe compared release dates and version names in our catalogue.
qwen/qwen3.7-max:thinking · thinking · 🇸🇬 SG
Inference by Alibaba Cloud
134.9
From 225 benchmarks. Plenty of data behind it.
112 ±5
Coding IQ 112.2, 95% interval 107–117. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$2.625cache read $0.525$7.875Not published by the provider
44.2
Poor value (44.2). Costs more per task than its benchmark results predict. Plenty of data behind it.
55 ±9
Typical (55 ±9). Generates about as many tokens as models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Claude Haiku 5.5
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
134.6
From 230 benchmarks. Plenty of data behind it.
124 ±6
Coding IQ 123.6, 95% interval 118–129. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.10cache read $0.01$0.50cache write $0.1251M
62.1
Good value (62.1). Cheaper per task than its benchmark results predict. Plenty of data behind it.
49 ±9
Typical (49 ±9). Generates about as many tokens as models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
160.3
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
MiniMax M3.1 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
134.5
From 60 benchmarks. Plenty of data behind it.
119 ±9
Coding IQ 118.7, 95% interval 109–128. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.15cache read $0.015$0.601M
60.2
Good value (60.2). Cheaper per task than its benchmark results predict. Plenty of data behind it.
Not measuredNot measured
MiniMax M3.1 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
🇺🇸 US
Inference by Fireworks AI
134.5
From 60 benchmarks. Plenty of data behind it.
119 ±9
Coding IQ 118.7, 95% interval 109–128. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0158$0.63Not published by the provider
60.2
Good value (60.2). Cheaper per task than its benchmark results predict. Plenty of data behind it.
Not measuredNot measured
MiniMax M3.1 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
fp4 · 🇺🇸 US
Inference by Modelrun
134.5
From 60 benchmarks. Plenty of data behind it.
119 ±9
Coding IQ 118.7, 95% interval 109–128. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0158$0.63Not published by the provider
60.2
Good value (60.2). Cheaper per task than its benchmark results predict. Plenty of data behind it.
Not measuredNot measured
MiniMax M3.1 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
fp4 · GLB
Inference by Nebius AI Studio
134.5
From 60 benchmarks. Plenty of data behind it.
119 ±9
Coding IQ 118.7, 95% interval 109–128. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0158$0.63Not published by the provider
60.2
Good value (60.2). Cheaper per task than its benchmark results predict. Plenty of data behind it.
Not measuredNot measured
MiniMax M3.1 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
🇺🇸 US
Inference by Novita AI
134.5
From 60 benchmarks. Plenty of data behind it.
119 ±9
Coding IQ 118.7, 95% interval 109–128. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0158$0.63Not published by the provider
60.2
Good value (60.2). Cheaper per task than its benchmark results predict. Plenty of data behind it.
Not measuredNot measured
MiniMax M3.1 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
fp8 · 🇺🇸 US
Inference by Parasail
134.5
From 60 benchmarks. Plenty of data behind it.
119 ±9
Coding IQ 118.7, 95% interval 109–128. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0158$0.63Not published by the provider
60.2
Good value (60.2). Cheaper per task than its benchmark results predict. Plenty of data behind it.
Not measuredNot measured
MiniMax M3.1 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
fp4 · UNK
Inference by Relace
134.5
From 60 benchmarks. Plenty of data behind it.
119 ±9
Coding IQ 118.7, 95% interval 109–128. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0945cache read $0.0105$0.525Not published by the provider
60.2
Good value (60.2). Cheaper per task than its benchmark results predict. Plenty of data behind it.
Not measuredNot measured
MiniMax M3.1 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
UNK
Inference by Runware
134.5
From 60 benchmarks. Plenty of data behind it.
119 ±9
Coding IQ 118.7, 95% interval 109–128. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0158$0.63Not published by the provider
60.2
Good value (60.2). Cheaper per task than its benchmark results predict. Plenty of data behind it.
Not measuredNot measured
MiniMax M3.1 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
fp8 · 🇨🇳 CN
Inference by Streamlake
134.5
From 60 benchmarks. Plenty of data behind it.
119 ±9
Coding IQ 118.7, 95% interval 109–128. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1418cache read $0.0142$0.567Not published by the provider
60.2
Good value (60.2). Cheaper per task than its benchmark results predict. Plenty of data behind it.
Not measuredNot measured
MiniMax M3.1 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
fp8 · 🇪🇺 EU
Inference by Tensorix
134.5
From 60 benchmarks. Plenty of data behind it.
119 ±9
Coding IQ 118.7, 95% interval 109–128. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.315cache read $0.0788$1.26Not published by the provider
60.2
Good value (60.2). Cheaper per task than its benchmark results predict. Plenty of data behind it.
Not measuredNot measured
MiniMax M3.1 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
🇺🇸 US
Inference by Together AI
134.5
From 60 benchmarks. Plenty of data behind it.
119 ±9
Coding IQ 118.7, 95% interval 109–128. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0158$0.63Not published by the provider
60.2
Good value (60.2). Cheaper per task than its benchmark results predict. Plenty of data behind it.
Not measuredNot measured
MiniMax M3.1 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
bf16 · GLB
Inference by Uomi
134.5
From 60 benchmarks. Plenty of data behind it.
119 ±9
Coding IQ 118.7, 95% interval 109–128. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1612$0.645Not published by the provider
60.2
Good value (60.2). Cheaper per task than its benchmark results predict. Plenty of data behind it.
Not measuredNot measured
MiniMax M3.1 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
nvfp4 · 🇺🇸 US
Inference by Wafer
134.5
From 60 benchmarks. Plenty of data behind it.
119 ±9
Coding IQ 118.7, 95% interval 109–128. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0158$0.63Not published by the provider
60.2
Good value (60.2). Cheaper per task than its benchmark results predict. Plenty of data behind it.
Not measuredNot measured
Gemini 3.6 Flash
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGemini 3.8 FlashWe compared release dates and version names in our catalogue.
134.4
From 267 benchmarks. Plenty of data behind it.
112 ±8
Coding IQ 111.8, 95% interval 103–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.75cache read $0.075$3.751M
50.6
Fair value (50.6). Costs per task about what its benchmark results predict. Plenty of data behind it.
59 ±10
Chatty (59 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
155.2
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
MiMo V2.6 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
134.3
From 154 benchmarks. Plenty of data behind it.
116 ±7
Coding IQ 115.7, 95% interval 109–123. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.14cache read $0.0028$0.281M
78.3
Good value (78.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
62 ±9
Chatty (62 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
Not measured
MiMo V2.6 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
fp8 · 🇺🇸 US
Inference by DeepInfra
134.3
From 154 benchmarks. Plenty of data behind it.
116 ±7
Coding IQ 115.7, 95% interval 109–123. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.147cache read $0.0029$0.294Not published by the provider
78.3
Good value (78.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
62 ±9
Chatty (62 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
Not measured
MiMo V2.6 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
nvfp4 · GLB
Inference by Heabsy
134.3
From 154 benchmarks. Plenty of data behind it.
116 ±7
Coding IQ 115.7, 95% interval 109–123. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.50cache read $0.1995$1.50Not published by the provider
78.3
Good value (78.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
62 ±9
Chatty (62 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
Not measured
MiMo V2.6 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
🇺🇸 US
Inference by Inworld
134.3
From 154 benchmarks. Plenty of data behind it.
116 ±7
Coding IQ 115.7, 95% interval 109–123. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.147cache read $0.0032$0.294Not published by the provider
78.3
Good value (78.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
62 ±9
Chatty (62 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
Not measured
MiMo V2.6 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
🇺🇸 US
Inference by Novita AI
134.3
From 154 benchmarks. Plenty of data behind it.
116 ±7
Coding IQ 115.7, 95% interval 109–123. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.147cache read $0.0029$0.294Not published by the provider
78.3
Good value (78.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
62 ±9
Chatty (62 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
Not measured
MiMo V2.6 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
UNK
Inference by Pareto
134.3
From 154 benchmarks. Plenty of data behind it.
116 ±7
Coding IQ 115.7, 95% interval 109–123. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.14cache read $0.0028$0.28Not published by the provider
78.3
Good value (78.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
62 ±9
Chatty (62 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
Not measured
MiMo V2.6 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
bf16 · GLB
Inference by Uomi
134.3
From 154 benchmarks. Plenty of data behind it.
116 ±7
Coding IQ 115.7, 95% interval 109–123. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1764$0.3528Not published by the provider
78.3
Good value (78.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
62 ±9
Chatty (62 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
Not measured
MiMo V2.6 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
fp8 · 🇺🇸 US
Inference by Venice
134.3
From 154 benchmarks. Plenty of data behind it.
116 ±7
Coding IQ 115.7, 95% interval 109–123. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1838cache read $0.0039$0.3675Not published by the provider
78.3
Good value (78.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
62 ±9
Chatty (62 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
Not measured
MiMo V2.6 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
fp8 · 🇨🇳 CN
Inference by Xiaomi
134.3
From 154 benchmarks. Plenty of data behind it.
116 ±7
Coding IQ 115.7, 95% interval 109–123. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.147cache read $0.0029$0.294Not published by the provider
78.3
Good value (78.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
62 ±9
Chatty (62 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
Not measured
Fugu Ultra
🇯🇵
Maker headquartered in JP, according to the company's own website. The citation is on the model page.
134.2
From 9 benchmarks. A fair amount of data behind it.
113 ±8
Coding IQ 113.4, 95% interval 106–121. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$5.00cache read $0.50$30.001MNot enough data yetNot measuredNot measured
Claude Opus 4.5
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
Newer version
Newer versionClaude Opus 5.5We compared release dates and version names in our catalogue.
134.0
From 384 benchmarks. Plenty of data behind it.
113 ±5
Coding IQ 112.8, 95% interval 108–118. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$5.00cache read $0.50$25.00cache write $6.25200K
46.0
Poor value (46.0). Costs more per task than its benchmark results predict. Plenty of data behind it.
43 ±10
Terse (43 ±10). Generates fewer tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
151.9
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Claude Opus 4.5
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
Newer version
Newer versionClaude Opus 5.5We compared release dates and version names in our catalogue.
anthropic/claude-opus-4.5:thinking · thinking
134.0
From 384 benchmarks. Plenty of data behind it.
113 ±5
Coding IQ 112.8, 95% interval 108–118. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$5.00cache read $0.50$25.00cache write $6.25200K
46.0
Poor value (46.0). Costs more per task than its benchmark results predict. Plenty of data behind it.
43 ±10
Terse (43 ±10). Generates fewer tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
151.9
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GPT-5.1 Codex Max
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
134.0
From 11 benchmarks. A fair amount of data behind it.
112 ±9
Coding IQ 111.7, 95% interval 103–121. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$2.50cache read $0.25$20.00400K
47.0
Poor value (47.0). Costs more per task than its benchmark results predict. A fair amount of data behind it.
48 ±9
Typical (48 ±9). Generates about as many tokens as models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
Not measured
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.53cache read $0.10$1.791M
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
tee
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.25cache read $0.125$3.951M
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
TEE/glm-5.2:thinking · tee+thinking
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.25cache read $0.125$3.951M
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.53cache read $0.10$1.791M
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
fp8 · 🇸🇬 SG
Inference by Alibaba Cloud
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.0143cache read $0.2029$3.1878Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · fp8 · 🇸🇬 SG
Inference by Alibaba Cloud
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.0143cache read $0.2029$3.1878Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
fp8
Inference by Ambient
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.63cache read $0.1575$2.10Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · fp8
Inference by Ambient
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.63cache read $0.1575$2.10Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
Inference by Aster
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.05cache read $0.21$4.20Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking
Inference by Aster
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.05cache read $0.21$4.20Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · fp8 · 🇺🇸 US
Inference by Atlascloud
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.47cache read $0.273$4.62Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
fp8 · 🇨🇳 CN
Inference by Baidu
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1985cache read $0.0369$0.6237Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · fp8 · 🇨🇳 CN
Inference by Baidu
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1985cache read $0.0369$0.6237Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
fp8 · 🇺🇸 US
Inference by Baseten
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.47cache read $0.147$4.62Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · fp8 · 🇺🇸 US
Inference by Baseten
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.47cache read $0.147$4.62Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
fp4 · 🇺🇸 US
Inference by Coreweave
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.651cache read $0.147$2.541Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · fp4 · 🇺🇸 US
Inference by Coreweave
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.651cache read $0.147$2.541Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
fp4 · 🇺🇸 US
Inference by DeepInfra
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.5906cache read $0.1103$1.89Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · fp4 · 🇺🇸 US
Inference by DeepInfra
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.5906cache read $0.1103$1.89Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
GLB
Inference by Engy
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.714cache read $0.189$1.575Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · GLB
Inference by Engy
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.714cache read $0.189$1.575Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
🇺🇸 US
Inference by Friendli AI
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.47cache read $0.273$4.62Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · 🇺🇸 US
Inference by Friendli AI
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.47cache read $0.273$4.62Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
GLB
Inference by Kitani
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.47cache read $0.273$4.62Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · GLB
Inference by Kitani
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.47cache read $0.273$4.62Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
fp8 · 🇺🇸 US
Inference by Meganova
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.05cache read $0.189$3.15Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · fp8 · 🇺🇸 US
Inference by Meganova
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.05cache read $0.189$3.15Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
fp8 · 🇺🇸 US
Inference by Morph
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.6752cache read $0.2279$2.0769Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · fp8 · 🇺🇸 US
Inference by Morph
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.6752cache read $0.2279$2.0769Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
fp4 · GLB
Inference by Nebius AI Studio
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.47cache read $0.147$4.62Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · fp4 · GLB
Inference by Nebius AI Studio
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.47cache read $0.147$4.62Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
fp8 · 🇺🇸 US
Inference by Novita AI
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.47cache read $0.273$4.62Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · fp8 · 🇺🇸 US
Inference by Novita AI
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.47cache read $0.273$4.62Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
UNK
Inference by Ollama
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.47cache read $0.273$4.62Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · UNK
Inference by Ollama
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.47cache read $0.273$4.62Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · fp4 · 🇺🇸 US
Inference by Parasail
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.155cache read $0.21$3.465Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
fp4 · UNK
Inference by Relace
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.3255cache read $0.0788$2.52Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · fp4 · UNK
Inference by Relace
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.3255cache read $0.0788$2.52Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
UNK
Inference by Runware
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.756cache read $0.1512$2.4098Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · UNK
Inference by Runware
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.756cache read $0.1512$2.4098Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
bf16 · 🇪🇺 EU
Inference by Scx Eu
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.6353cache read $0.1617$2.2292Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · bf16 · 🇪🇺 EU
Inference by Scx Eu
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.6353cache read $0.1617$2.2292Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
bf16 · GLB
Inference by Scx
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.6353cache read $0.1617$2.2292Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · bf16 · GLB
Inference by Scx
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.6353cache read $0.1617$2.2292Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
fp8 · 🇸🇬 SG
Inference by SiliconFlow
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.6836cache read $0.1365$2.1483Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · fp8 · 🇸🇬 SG
Inference by SiliconFlow
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.6836cache read $0.1365$2.1483Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
fp8 · 🇨🇳 CN
Inference by Streamlake
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.294cache read $0.0546$0.924Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · fp8 · 🇨🇳 CN
Inference by Streamlake
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.294cache read $0.0546$0.924Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
fp8 · 🇪🇺 EU
Inference by Tensorix
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.155cache read $0.2888$3.465Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · fp8 · 🇪🇺 EU
Inference by Tensorix
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.155cache read $0.2888$3.465Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
🇺🇸 US
Inference by Together AI
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.47cache read $0.273$4.62Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · 🇺🇸 US
Inference by Together AI
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.47cache read $0.273$4.62Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
bf16 · GLB
Inference by Uomi
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.2737$4.0082Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · bf16 · GLB
Inference by Uomi
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.2737$4.0082Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
fp8 · 🇺🇸 US
Inference by Venice
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.47cache read $0.273$4.62Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · fp8 · 🇺🇸 US
Inference by Venice
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.47cache read $0.273$4.62Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
fp4 · 🇺🇸 US
Inference by Wafer Us
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0074cache read $0.0063$7.14Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · fp4 · 🇺🇸 US
Inference by Wafer Us
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0074cache read $0.0063$7.14Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
fp4 · 🇺🇸 US
Inference by Wafer
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1995cache read $0.189$8.40Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · fp4 · 🇺🇸 US
Inference by Wafer
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1995cache read $0.189$8.40Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
fp8 · 🇸🇬 SG
Inference by Z.ai
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.47cache read $0.273$4.62Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.2
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGLM 5.3We compared release dates and version names in our catalogue.
z-ai/glm-5.2:thinking · thinking · fp8 · 🇸🇬 SG
Inference by Z.ai
133.9
From 361 benchmarks. Plenty of data behind it.
113 ±9
Coding IQ 113.3, 95% interval 105–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.47cache read $0.273$4.62Not published by the provider
45.9
Poor value (45.9). Costs more per task than its benchmark results predict. Plenty of data behind it.
58 ±10
Chatty (58 ±10). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
153.1
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Claude Sonnet 4.6
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
Newer version
Newer versionClaude Sonnet 5.5We compared release dates and version names in our catalogue.
133.8
From 469 benchmarks. Plenty of data behind it.
108 ±6
Coding IQ 108.4, 95% interval 103–114. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$3.00cache read $0.30$15.00cache write $3.751M
35.4
Poor value (35.4). Costs more per task than its benchmark results predict. Plenty of data behind it.
71 ±8
Chatty (71 ±8). Generates more tokens than models with similar coding results. Measured at default effort. A fair amount of data behind it.
152.9
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Claude Sonnet 4.6
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
Newer version
Newer versionClaude Sonnet 5.5We compared release dates and version names in our catalogue.
anthropic/claude-sonnet-4.6:thinking · thinking
133.8
From 469 benchmarks. Plenty of data behind it.
108 ±6
Coding IQ 108.4, 95% interval 103–114. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$3.00cache read $0.30$15.00cache write $3.751M
35.4
Poor value (35.4). Costs more per task than its benchmark results predict. Plenty of data behind it.
71 ±8
Chatty (71 ±8). Generates more tokens than models with similar coding results. Measured at default effort. A fair amount of data behind it.
152.9
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.10cache read $0.025$0.301M
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
tee
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.15cache read $0.03$0.501M
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
NVFP4 · 🇰🇷 KR
Inference by Aoru
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0788$0.525Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
nvfp4 · 🇮🇳 IN
Inference by Astorias
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0473cache read $0.0315$0.231Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
fp8 · 🇨🇳 CN
Inference by Baidu
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0315$0.525Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
fp8 · 🇺🇸 US
Inference by Baseten
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0315$0.525Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
nvfp4 · 🇺🇸 US
Inference by Coreweave
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.126cache read $0.021$0.399Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
fp4 · 🇺🇸 US
Inference by Crusoe
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0315$0.525Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
fp4 · GLB
Inference by Darkbloom
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1365cache read $0.0341$0.378Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
fp4 · 🇺🇸 US
Inference by DeepInfra
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1055cache read $0.0211$0.3518Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
nvfp4 · 🇪🇺 EU
Inference by Discretestack
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.10cache read $0.025$0.30Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
GLB
Inference by Engy
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1418cache read $0.0284$0.4725Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
nvfp4 · 🇪🇺 EU
Inference by Entrim
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.105cache read $0.0184$0.3675Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
🇺🇸 US
Inference by Fireworks AI
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0315$0.525Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
🇺🇸 US
Inference by Friendli AI
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0315$0.525Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
fp8 · 🇪🇺 EU
Inference by Greenference
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0263$0.0735Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
fp8 · 🇪🇺 EU
Inference by Inceptron
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.168cache read $0.042$0.525Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
GLB
Inference by Kitani
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0315$0.525Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
fp4 · 🇺🇸 US
Inference by Modal
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0315$0.525Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
fp8 · 🇺🇸 US
Inference by Morph
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1061cache read $0.0525$0.3045Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
nvfp4 · 🇪🇺 EU
Inference by Munito
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0315$0.525Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
fp4 · GLB
Inference by Nebius AI Studio
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0315$0.525Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
fp8 · 🇺🇸 US
Inference by Neuralwatt
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0315$0.525Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
flex · fp8 · 🇺🇸 US
Inference by Neuralwatt
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1024cache read $0.0205$0.3413Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
fp8 · 🇺🇸 US
Inference by Novita AI
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0882cache read $0.0176$0.294Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Inference by Nube
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0866cache read $0.0173$0.2888Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
UNK
Inference by Ollama
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0315$0.525Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
🇸🇬 SG
Inference by Openscale
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.126cache read $0.0252$0.42Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
fp8 · 🇺🇸 US
Inference by Parasail
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0315$0.525Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
fast · fp4 · 🇺🇸 US
Inference by Parasail
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1969cache read $0.0394$0.6563Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
UNK
Inference by Pareto
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0945cache read $0.0189$0.315Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
🇺🇸 US
Inference by Pokee
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0578$0.21Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
tee · GLB
Inference by Redpill
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0315$0.525Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
UNK
Inference by Reka AI
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.10cache read $0.025$0.30Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
fp4 · UNK
Inference by Relace
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0419cache read $0.0315$0.0823Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
FI/IS
Inference by River
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.126cache read $0.0252$0.42Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
fp8 · 🇺🇸 US
Inference by Runanywhere
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0315$0.525Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Inference by Runinfra
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0536cache read $0.0515$0.4043Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
UNK
Inference by Runware
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1418cache read $0.0284$0.4725Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
standard · fp4 · GLB
Inference by Sail Research
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1838cache read $0.0236$0.6563Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
fp8 · GLB
Inference by Scx
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0924cache read $0.0263$0.2625Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
fp8 · 🇸🇬 SG
Inference by SiliconFlow
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.105cache read $0.0315$0.42Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
fp8 · 🇨🇳 CN
Inference by Streamlake
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0929cache read $0.0186$0.3098Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
🇪🇺 EU
Inference by Tensorix
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.21cache read $0.0525$0.525Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
🇺🇸 US
Inference by Together AI
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0315$0.525Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
bf16 · GLB
Inference by Uomi
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1291$0.4525Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
🇺🇸 US
Inference by Venice
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0315$0.525Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
nvfp4 · 🇺🇸 US
Inference by Wafer
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0525cache read $0.0315$0.21Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.3 Flash
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
fp8 · 🇸🇬 SG
Inference by Z.ai
133.6
From 266 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0315$0.525Not published by the provider
65.3
Good value (65.3). Cheaper per task than its benchmark results predict. Plenty of data behind it.
68 ±11
Chatty (68 ±11). Generates more tokens than models with similar coding results. Measured at max effort. A fair amount of data behind it.
153.8
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GPT-5.3 Codex
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
133.6
From 51 benchmarks. Plenty of data behind it.
114 ±8
Coding IQ 113.7, 95% interval 106–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.75cache read $0.175$14.00400K
59.7
Good value (59.7). Cheaper per task than its benchmark results predict. Plenty of data behind it.
46 ±9
Typical (46 ±9). Generates about as many tokens as models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
Not measured
GPT 5.2
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGPT 6We compared the version names in our catalogue.
133.4
From 440 benchmarks. Plenty of data behind it.
112 ±7
Coding IQ 111.5, 95% interval 105–118. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.75cache read $0.175$14.00400K
56.6
Good value (56.6). Cheaper per task than its benchmark results predict. Plenty of data behind it.
49 ±14
Typical (49 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.5
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
≈ 133
From 3 benchmarks. Little data behind it, so treat it as a rough estimate.
Not enough data yet$3.00cache read $0.75$6.00262KNot enough data yetNot measuredNot measured
Inference by Aionlabs
≈ 133
From 3 benchmarks. Little data behind it, so treat it as a rough estimate.
Not enough data yet$3.15cache read $0.7875$6.30Not published by the providerNot enough data yetNot measuredNot measured
GPT-5 Pro
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGPT-5.5 ProWe compared release dates and version names in our catalogue.
132.8
From 14 benchmarks. Plenty of data behind it.
Not enough data yet$15.00cache read $1.50$120.00400K
29.6
Poor value (29.6). Costs more per task than its benchmark results predict. A fair amount of data behind it.
Not measured
151.3
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Step 5 Preview
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
132.7
From 119 benchmarks. Plenty of data behind it.
115 ±7
Coding IQ 115.2, 95% interval 108–123. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.00cache read $0.05$2.701M
43.2
Poor value (43.2). Costs more per task than its benchmark results predict. Plenty of data behind it.
Not measuredNot measured
Step 5 Preview
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
fp8 · 🇺🇸 US
Inference by DeepInfra
132.7
From 119 benchmarks. Plenty of data behind it.
115 ±7
Coding IQ 115.2, 95% interval 108–123. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.05cache read $0.0525$2.835Not published by the provider
43.2
Poor value (43.2). Costs more per task than its benchmark results predict. Plenty of data behind it.
Not measuredNot measured
Step 5 Preview
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
🇺🇸 US
Inference by Novita AI
132.7
From 119 benchmarks. Plenty of data behind it.
115 ±7
Coding IQ 115.2, 95% interval 108–123. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.05cache read $0.105$2.835Not published by the provider
43.2
Poor value (43.2). Costs more per task than its benchmark results predict. Plenty of data behind it.
Not measuredNot measured
Step 5 Preview
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
bf16 · GLB
Inference by Uomi
132.7
From 119 benchmarks. Plenty of data behind it.
115 ±7
Coding IQ 115.2, 95% interval 108–123. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.1025$2.9768Not published by the provider
43.2
Poor value (43.2). Costs more per task than its benchmark results predict. Plenty of data behind it.
Not measuredNot measured
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.50cache read $0.125$2.60262K
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
tee
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.39cache read $0.039$1.80262K
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
moonshotai/kimi-k2.6:thinking · thinking
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.50cache read $0.125$2.60262K
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
fp4 · 🇨🇳 CN
Inference by Baidu
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.409cache read $0.0689$1.722Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
moonshotai/kimi-k2.6:thinking · thinking · fp4 · 🇨🇳 CN
Inference by Baidu
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.409cache read $0.0689$1.722Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
int4 · 🇺🇸 US
Inference by Chutes
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.4095cache read $0.041$1.89Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
moonshotai/kimi-k2.6:thinking · thinking · int4 · 🇺🇸 US
Inference by Chutes
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.4095cache read $0.041$1.89Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
fp4 · 🇺🇸 US
Inference by Coreweave
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.6825cache read $0.1575$3.5805Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
moonshotai/kimi-k2.6:thinking · thinking · fp4 · 🇺🇸 US
Inference by Coreweave
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.6825cache read $0.1575$3.5805Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
bf16 · 🇺🇸 US
Inference by Crusoe
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.735cache read $0.3675$3.675Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
moonshotai/kimi-k2.6:thinking · thinking · bf16 · 🇺🇸 US
Inference by Crusoe
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.735cache read $0.3675$3.675Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
fp4 · 🇺🇸 US
Inference by DeepInfra
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.7875cache read $0.1575$3.675Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
int4 · 🇪🇺 EU
Inference by Inceptron
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.593cache read $0.1995$2.9274Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
moonshotai/kimi-k2.6:thinking · thinking · int4 · 🇪🇺 EU
Inference by Inceptron
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.593cache read $0.1995$2.9274Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
GLB
Inference by Kitani
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.9975cache read $0.168$4.20Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
moonshotai/kimi-k2.6:thinking · thinking · GLB
Inference by Kitani
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.9975cache read $0.168$4.20Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
🇸🇬 SG
Inference by Moonshot AI
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.9975cache read $0.168$4.20Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
moonshotai/kimi-k2.6:thinking · thinking · 🇸🇬 SG
Inference by Moonshot AI
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.9975cache read $0.168$4.20Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
🇺🇸 US
Inference by Novita AI
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.84cache read $0.168$3.57Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
moonshotai/kimi-k2.6:thinking · thinking · 🇺🇸 US
Inference by Novita AI
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.84cache read $0.168$3.57Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
UNK
Inference by Ollama
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.9975cache read $0.168$4.20Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
moonshotai/kimi-k2.6:thinking · thinking · UNK
Inference by Ollama
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.9975cache read $0.168$4.20Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
int4 · 🇺🇸 US
Inference by Parasail
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.7875cache read $0.168$3.675Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
moonshotai/kimi-k2.6:thinking · thinking · int4 · 🇺🇸 US
Inference by Parasail
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.7875cache read $0.168$3.675Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
asap · fp8 · GLB
Inference by Sail Research
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.05cache read $0.21$4.20Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
moonshotai/kimi-k2.6:thinking · asap+thinking · fp8 · GLB
Inference by Sail Research
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.05cache read $0.21$4.20Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
fp8 · 🇨🇳 CN
Inference by Streamlake
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.6284cache read $0.1058$2.646Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Kimi K2.6
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionKimi K3.1We compared the version names in our catalogue.
moonshotai/kimi-k2.6:thinking · thinking · fp8 · 🇨🇳 CN
Inference by Streamlake
132.6
From 305 benchmarks. Plenty of data behind it.
113 ±7
Coding IQ 112.9, 95% interval 106–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.6284cache read $0.1058$2.646Not published by the provider
50.4
Fair value (50.4). Costs per task about what its benchmark results predict. Plenty of data behind it.
48 ±14
Typical (48 ±14). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
151.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.044cache read $0.0014$0.1321M
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.044cache read $0.0014$0.1321M
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
🇸🇬 SG
Inference by Alibaba Cloud
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.3696cache read $0.037$1.1088Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking · 🇸🇬 SG
Inference by Alibaba Cloud
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.3696cache read $0.037$1.1088Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
fp8 · 🇨🇳 CN
Inference by Baidu
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0462cache read $0.0015$0.1386Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking · fp8 · 🇨🇳 CN
Inference by Baidu
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0462cache read $0.0015$0.1386Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
fp8 · 🇺🇸 US
Inference by Baseten
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1365cache read $0.0294$0.273Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking · fp8 · 🇺🇸 US
Inference by Baseten
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1365cache read $0.0294$0.273Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
fp8 · 🇺🇸 US
Inference by Chutes
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.462cache read $0.0462$1.386Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking · fp8 · 🇺🇸 US
Inference by Chutes
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.462cache read $0.0462$1.386Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
fp8 · 🇺🇸 US
Inference by Coreweave
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1365cache read $0.0525$0.2835Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking · fp8 · 🇺🇸 US
Inference by Coreweave
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1365cache read $0.0525$0.2835Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
fp8 · 🇺🇸 US
Inference by DeepInfra
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.063cache read $0.0158$0.189Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking · fp8 · 🇺🇸 US
Inference by DeepInfra
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.063cache read $0.0158$0.189Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
GLB
Inference by Engy
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0473cache read $0.0095$0.0945Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking · GLB
Inference by Engy
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0473cache read $0.0095$0.0945Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
GLB
Inference by Gonkaai
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0536$0.1092Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking · GLB
Inference by Gonkaai
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0536$0.1092Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
fp8 · 🇦🇪 AE
Inference by Hyperfusion
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0525$0.168Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking · fp8 · 🇦🇪 AE
Inference by Hyperfusion
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0525$0.168Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
fp4 · 🇪🇺 EU
Inference by Inceptron
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1365cache read $0.0284$0.5198Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking · fp4 · 🇪🇺 EU
Inference by Inceptron
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1365cache read $0.0284$0.5198Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
fp8
Inference by Mancer
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0525$0.525Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking · fp8
Inference by Mancer
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0525$0.525Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
mxfp4 · 🇪🇺 EU
Inference by Munito
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1155cache read $0.0231$0.294Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking · mxfp4 · 🇪🇺 EU
Inference by Munito
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1155cache read $0.0231$0.294Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
fp8 · 🇺🇸 US
Inference by Novita AI
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.4389cache read $0.0279$1.3167Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking · fp8 · 🇺🇸 US
Inference by Novita AI
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.4389cache read $0.0279$1.3167Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
UNK
Inference by Ollama
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.462cache read $0.0147$1.386Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking · UNK
Inference by Ollama
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.462cache read $0.0147$1.386Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
fp8 · 🇺🇸 US
Inference by Parasail
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.147cache read $0.0525$0.294Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking · fp8 · 🇺🇸 US
Inference by Parasail
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.147cache read $0.0525$0.294Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
fp4 · UNK
Inference by Primeintellect
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.462cache read $0.0147$1.386Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking · fp4 · UNK
Inference by Primeintellect
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.462cache read $0.0147$1.386Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
fp4 · UNK
Inference by Relace
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0735cache read $0.0147$0.147Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking · fp4 · UNK
Inference by Relace
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0735cache read $0.0147$0.147Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
fp8 · 🇺🇸 US
Inference by Runanywhere
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.3077cache read $0.0098$0.924Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking · fp8 · 🇺🇸 US
Inference by Runanywhere
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.3077cache read $0.0098$0.924Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
standard · fp4 · GLB
Inference by Sail Research
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0441$0.315Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · standard+thinking · fp4 · GLB
Inference by Sail Research
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0441$0.315Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
fp8 · 🇸🇬 SG
Inference by SiliconFlow
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.231cache read $0.0294$0.693Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking · fp8 · 🇸🇬 SG
Inference by SiliconFlow
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.231cache read $0.0294$0.693Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
fp8 · 🇨🇳 CN
Inference by Streamlake
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0924cache read $0.0029$0.2772Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking · fp8 · 🇨🇳 CN
Inference by Streamlake
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.0924cache read $0.0029$0.2772Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
fp4 · 🇪🇺 EU
Inference by Tensorix
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0394$0.315Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking · fp4 · 🇪🇺 EU
Inference by Tensorix
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1575cache read $0.0394$0.315Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
bf16 · GLB
Inference by Uomi
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1739$0.4752Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking · bf16 · GLB
Inference by Uomi
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1739$0.4752Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
🇺🇸 US
Inference by Venice
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1365cache read $0.0294$0.273Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking · 🇺🇸 US
Inference by Venice
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1365cache read $0.0294$0.273Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
nvfp4 · 🇺🇸 US
Inference by Wafer
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1365cache read $0.1313$0.2415Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Flash 0731
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
Newer version
Newer versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deepseek/deepseek-v4-flash-0731:thinking · thinking · nvfp4 · 🇺🇸 US
Inference by Wafer
132.2
From 149 benchmarks. Plenty of data behind it.
113 ±8
Coding IQ 112.8, 95% interval 105–121. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1365cache read $0.1313$0.2415Not published by the provider
64.8
Good value (64.8). Cheaper per task than its benchmark results predict. Plenty of data behind it.
57 ±9
Chatty (57 ±9). Generates more tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
154.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Gemini 3 Flash
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGemini 3.8 FlashWe compared release dates and version names in our catalogue.
google/gemini-3-flash-preview
132.0
From 439 benchmarks. Plenty of data behind it.
106 ±6
Coding IQ 106.1, 95% interval 100–112. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.50cache read $0.05$3.001M
62.2
Good value (62.2). Cheaper per task than its benchmark results predict. Plenty of data behind it.
53 ±15
Typical (53 ±15). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
152.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Gemini 3 Flash
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGemini 3.8 FlashWe compared release dates and version names in our catalogue.
google/gemini-3-flash-preview-thinking · thinking
132.0
From 439 benchmarks. Plenty of data behind it.
106 ±6
Coding IQ 106.1, 95% interval 100–112. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.50cache read $0.05$3.001M
62.2
Good value (62.2). Cheaper per task than its benchmark results predict. Plenty of data behind it.
53 ±15
Typical (53 ±15). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
152.4
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API

Data credits: Epoch AI Benchmarking Hub, CC BY 4.0 (licence)

How this data works