Skip to content

Listings

All listings: Aihubmix

Every listing for the models on page 4 of 20 of Aihubmix, including each model's other versions and routes. Back to Aihubmix

Sold by Aihubmix

45 listings on this page sold by Aihubmix.

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
GPT 5.1
🇺🇸
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.
131.0
From 352 benchmarks. Plenty of data behind it.
107 ±8
Coding IQ 106.6, 95% interval 99–114. 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$10.00400K
49.7
Fair value (49.7). Costs per task about what its benchmark results predict. Plenty of data behind it.
50 ±9
Typical (50 ±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.
150.9
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.1
🇨🇳
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 versionGLM 5.3We compared release dates and version names in our catalogue.
130.8
From 224 benchmarks. Plenty of data behind it.
110 ±6
Coding IQ 110.0, 95% interval 104–116. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.845cache read $0.1831$3.38200K
43.2
Poor value (43.2). Costs more per task than its benchmark results predict. Plenty of data behind it.
51 ±8
Typical (51 ±8). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
150.7
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.1
🇨🇳
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 versionGLM 5.3We compared release dates and version names in our catalogue.
cc-glm-5.1 · cc
130.8
From 224 benchmarks. Plenty of data behind it.
110 ±6
Coding IQ 110.0, 95% interval 104–116. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.06$0.22200K
43.2
Poor value (43.2). Costs more per task than its benchmark results predict. Plenty of data behind it.
51 ±8
Typical (51 ±8). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
150.7
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.1
🇨🇳
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 versionGLM 5.3We compared release dates and version names in our catalogue.
coding-glm-5.1 · coding
130.8
From 224 benchmarks. Plenty of data behind it.
110 ±6
Coding IQ 110.0, 95% interval 104–116. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.06$0.22200K
43.2
Poor value (43.2). Costs more per task than its benchmark results predict. Plenty of data behind it.
51 ±8
Typical (51 ±8). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
150.7
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.1
🇨🇳
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 versionGLM 5.3We compared release dates and version names in our catalogue.
coding-glm-5.1-free · coding+free
130.8
From 224 benchmarks. Plenty of data behind it.
110 ±6
Coding IQ 110.0, 95% interval 104–116. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.00$0.00200K
43.2
Poor value (43.2). Costs more per task than its benchmark results predict. Plenty of data behind it.
51 ±8
Typical (51 ±8). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
150.7
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.1
🇨🇳
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 versionGLM 5.3We compared release dates and version names in our catalogue.
alicloud-glm-5.1
Inference by Alibaba Cloud
130.8
From 224 benchmarks. Plenty of data behind it.
110 ±6
Coding IQ 110.0, 95% interval 104–116. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.84cache read $0.169$3.38cache write $1.0563200K
43.2
Poor value (43.2). Costs more per task than its benchmark results predict. Plenty of data behind it.
51 ±8
Typical (51 ±8). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
150.7
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5.1
🇨🇳
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 versionGLM 5.3We compared release dates and version names in our catalogue.
zai-glm-5.1
Inference by Z.ai
130.8
From 224 benchmarks. Plenty of data behind it.
110 ±6
Coding IQ 110.0, 95% interval 104–116. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.845cache read $0.1831$3.38200K
43.2
Poor value (43.2). Costs more per task than its benchmark results predict. Plenty of data behind it.
51 ±8
Typical (51 ±8). Generates about as many tokens as models with similar coding results. Measured at default effort. A fair amount of data behind it.
150.7
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Qwen3.6 Max Preview
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionCoding Qwen4 Max PreviewWe compared the version names in our catalogue.
130.8
From 99 benchmarks. Plenty of data behind it.
≈ 109 ±8
Coding IQ 109.3, 95% interval 102–117. Little data behind it, so treat it as a rough estimate. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.268cache read $0.1268$7.608cache write $1.585262K
41.5
Poor value (41.5). Costs more per task than its benchmark results predict. A fair amount of data behind it.
Not measured
149.9
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Pro
🇨🇳
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.
130.7
From 303 benchmarks. Plenty of data behind it.
105 ±7
Coding IQ 104.6, 95% interval 98–111. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.69cache read $0.1403$3.381M
46.9
Poor value (46.9). 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. A fair amount of data behind it.
150.2
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Pro
🇨🇳
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.
alicloud-deepseek-v4-pro
Inference by Alibaba Cloud
130.7
From 303 benchmarks. Plenty of data behind it.
105 ±7
Coding IQ 104.6, 95% interval 98–111. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.69cache read $0.13$3.381M
46.9
Poor value (46.9). 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. A fair amount of data behind it.
150.2
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 Pro
🇨🇳
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.
deep-deepseek-v4-pro
Inference by DeepSeek
130.7
From 303 benchmarks. Plenty of data behind it.
105 ±7
Coding IQ 104.6, 95% interval 98–111. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.478cache read $0.0043$0.9561M
46.9
Poor value (46.9). 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. A fair amount of data behind it.
150.2
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Qwen3.5 397B A17B
🇨🇳
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.
130.6
From 367 benchmarks. Plenty of data behind it.
105 ±7
Coding IQ 104.8, 95% interval 98–112. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.1644$0.9864262K
48.4
Fair value (48.4). Costs per task about what 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.
148
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Laguna S 2.1
🇺🇸
Maker headquartered in US, 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.
130.0
From 6 benchmarks. A fair amount of data behind it.
Not enough data yet$0.10cache read $0.01$0.201MNot enough data yetNot measuredNot measured
O3 Pro
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
130.0
From 28 benchmarks. Plenty of data behind it.
Not enough data yet$20.00$80.00200K
36.5
Poor value (36.5). Costs more per task than its benchmark results predict. A fair amount of data behind it.
Not measured
147.5
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Claude Sonnet 4.5
🇺🇸
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.
129.7
From 442 benchmarks. Plenty of data behind it.
107 ±7
Coding IQ 107.2, 95% interval 101–114. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$3.30cache read $0.33$16.50cache write $4.125200K
40.2
Poor value (40.2). Costs more per task than its benchmark results predict. Plenty of data behind it.
41 ±8
Terse (41 ±8). Generates fewer tokens than models with similar coding results. Measured at default effort. A fair amount of data behind it.
147.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Claude Sonnet 4.5
🇺🇸
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.
claude-sonnet-4-5-think · think
129.7
From 442 benchmarks. Plenty of data behind it.
107 ±7
Coding IQ 107.2, 95% interval 101–114. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$3.30cache read $0.33$16.50cache write $4.125200K
40.2
Poor value (40.2). Costs more per task than its benchmark results predict. Plenty of data behind it.
41 ±8
Terse (41 ±8). Generates fewer tokens than models with similar coding results. Measured at default effort. A fair amount of data behind it.
147.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5
🇨🇳
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.
alicloud-glm-5
129.7
From 306 benchmarks. Plenty of data behind it.
107 ±5
Coding IQ 106.7, 95% interval 102–112. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.5634cache read $0.1127$2.5353Not published by the provider
49.7
Fair value (49.7). Costs per task about what 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.
146.5
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5
🇨🇳
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.
cc-glm-5 · cc
129.7
From 306 benchmarks. Plenty of data behind it.
107 ±5
Coding IQ 106.7, 95% interval 102–112. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.06$0.22200K
49.7
Fair value (49.7). Costs per task about what 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.
146.5
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5
🇨🇳
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.
coding-glm-5 · coding
129.7
From 306 benchmarks. Plenty of data behind it.
107 ±5
Coding IQ 106.7, 95% interval 102–112. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.06$0.22200K
49.7
Fair value (49.7). Costs per task about what 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.
146.5
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
GLM 5
🇨🇳
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.
coding-glm-5-free · coding+free
129.7
From 306 benchmarks. Plenty of data behind it.
107 ±5
Coding IQ 106.7, 95% interval 102–112. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.00$0.00200K
49.7
Fair value (49.7). Costs per task about what 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.
146.5
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Grok 4.20 Multi-Agent
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
grok-4.20-multi-agent-0309 · 0309
≈ 130
From 73 benchmarks. Little data behind it, so treat it as a rough estimate.
Not enough data yet$2.00cache read $0.20$6.001MNot enough data yetNot measuredNot measured
Grok 4.20 Multi-Agent
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
grok-4.20-multi-agent-beta-0309 · beta+0309
≈ 130
From 73 benchmarks. Little data behind it, so treat it as a rough estimate.
Not enough data yet$2.00cache read $0.20$6.002MNot enough data yetNot measuredNot measured
Agents A1
Open weights. The weights are publicly available, according to the provider's model list or a public Hugging Face repo.
agents-a1-free · free
≈ 130
From 4 benchmarks. Little data behind it, so treat it as a rough estimate.
Not enough data yet$0.00$0.00262KNot enough data yetNot measuredNot measured
129.6
From 64 benchmarks. Plenty of data behind it.
104 ±11
Coding IQ 104.2, 95% interval 94–115. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.9295cache read $0.1859$4.6475256KNot enough data yetNot measuredNot measured
Doubao Seed 2.1 Prodoubao-seed-2-1-pro-260628 · 260628
129.6
From 64 benchmarks. Plenty of data behind it.
104 ±11
Coding IQ 104.2, 95% interval 94–115. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.9295cache read $0.1859$4.6475256KNot enough data yetNot measuredNot measured
Mistral Large 4
🇫🇷
Maker headquartered in FR, according to the company's own website. The citation is on the model page.
mistral-large-4-0
129.6
From 173 benchmarks. Plenty of data behind it.
115 ±7
Coding IQ 114.7, 95% interval 108–122. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.36cache read $0.1399$4.181M
30.8
Poor value (30.8). Costs more 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.
154.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Qwen3.7 Plus
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
129.5
From 260 benchmarks. Plenty of data behind it.
109 ±6
Coding IQ 108.7, 95% interval 103–115. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.282cache read $0.0564$1.128cache write $0.35251M
47.1
Poor value (47.1). Costs more per task than its benchmark results predict. Plenty of data behind it.
59 ±9
Chatty (59 ±9). Generates more tokens than models with similar coding results. Measured with reasoning off. Little data behind it, so treat it as a rough estimate.
147.9
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Qwen3.6 Plus
🇨🇳
Maker headquartered in CN, according to the company's own website. The citation is on the model page.
Newer version
Newer versionQwen3.7 PlusWe compared release dates and version names in our catalogue.
129.4
From 256 benchmarks. Plenty of data behind it.
106 ±5
Coding IQ 105.7, 95% interval 101–110. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.282cache read $0.0282$1.692cache write $0.35251M
38.6
Poor value (38.6). Costs more per task than its benchmark results predict. Plenty of data behind it.
44 ±12
Terse (44 ±12). Generates fewer tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
148.9
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Qwen 3.8 27B
🇨🇳
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.
129.4
From 251 benchmarks. Plenty of data behind it.
111 ±7
Coding IQ 111.4, 95% interval 104–119. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.20cache read $0.05$2.501M
39.7
Poor value (39.7). Costs more per task than its benchmark results predict. Plenty of data behind it.
Not measured
152.6
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Grok 4.20
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
grok-4.20-beta-0309-non-reasoning · beta+non-reasoning+0309
129.3
From 281 benchmarks. Plenty of data behind it.
108 ±8
Coding IQ 108.2, 95% interval 101–116. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$2.00cache read $1.00$6.002M
40.6
Poor value (40.6). Costs more per task than its benchmark results predict. Plenty of data behind it.
45 ±9
Typical (45 ±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.
152.3
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Grok 4.20
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
grok-4.20-beta-0309-reasoning · beta+reasoning+0309
129.3
From 281 benchmarks. Plenty of data behind it.
108 ±8
Coding IQ 108.2, 95% interval 101–116. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$2.00cache read $1.00$6.002M
40.6
Poor value (40.6). Costs more per task than its benchmark results predict. Plenty of data behind it.
45 ±9
Typical (45 ±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.
152.3
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Grok 4.20
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
grok-4-20-non-reasoning · non-reasoning
129.3
From 281 benchmarks. Plenty of data behind it.
108 ±8
Coding IQ 108.2, 95% interval 101–116. 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$6.001M
40.6
Poor value (40.6). Costs more per task than its benchmark results predict. Plenty of data behind it.
45 ±9
Typical (45 ±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.
152.3
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Grok 4.20
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
grok-4-20-reasoning · reasoning
129.3
From 281 benchmarks. Plenty of data behind it.
108 ±8
Coding IQ 108.2, 95% interval 101–116. 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$6.001M
40.6
Poor value (40.6). Costs more per task than its benchmark results predict. Plenty of data behind it.
45 ±9
Typical (45 ±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.
152.3
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 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 versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
129.1
From 229 benchmarks. Plenty of data behind it.
108 ±12
Coding IQ 107.8, 95% interval 95–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.142cache read $0.0284$0.2841M
74.6
Good value (74.6). Cheaper per task than its benchmark results predict. Plenty of data behind it.
44 ±9
Terse (44 ±9). Generates fewer tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
146.9
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 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 versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
alicloud-deepseek-v4-flash
Inference by Alibaba Cloud
129.1
From 229 benchmarks. Plenty of data behind it.
108 ±12
Coding IQ 107.8, 95% interval 95–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.14cache read $0.028$0.281M
74.6
Good value (74.6). Cheaper per task than its benchmark results predict. Plenty of data behind it.
44 ±9
Terse (44 ±9). Generates fewer tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
146.9
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
DeepSeek V4 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 versionDeepSeek V4.1 Flash 1017We compared release dates and version names in our catalogue.
deep-deepseek-v4-flash
Inference by DeepSeek
129.1
From 229 benchmarks. Plenty of data behind it.
108 ±12
Coding IQ 107.8, 95% interval 95–120. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.154cache read $0.0308$0.3081M
74.6
Good value (74.6). Cheaper per task than its benchmark results predict. Plenty of data behind it.
44 ±9
Terse (44 ±9). Generates fewer tokens than models with similar coding results. Measured at default effort. Little data behind it, so treat it as a rough estimate.
146.9
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Doubao Seed 2.0 Pro
Newer version
Newer versionDoubao Seed 2.1 ProWe compared release dates and version names in our catalogue.
129.0
From 86 benchmarks. Plenty of data behind it.
≈ 102 ±10
Coding IQ 101.6, 95% interval 92–112. Little data behind it, so treat it as a rough estimate. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$0.4958cache read $0.0992$2.479256KNot enough data yetNot measuredNot measured
GPT-5.1 Codex
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
Newer version
Newer versionGPT-5.3 CodexWe compared the version names in our catalogue.
129.0
From 31 benchmarks. Plenty of data behind it.
106 ±7
Coding IQ 106.2, 95% interval 99–113. A fair amount 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$10.00400K
47.0
Poor value (47.0). Costs more per task than its benchmark results predict. A fair amount of data behind it.
54 ±9
Typical (54 ±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
Gemini 2.5 Pro
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
128.8
From 374 benchmarks. Plenty of data behind it.
101 ±18
Coding IQ 100.6, 95% interval 83–118. 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$10.001M
36.3
Poor value (36.3). Costs more per task than its benchmark results predict. Plenty of data behind it.
54 ±9
Typical (54 ±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.
144.9
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Gemini 2.5 Pro
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
gemini-2.5-pro-search · search
128.8
From 374 benchmarks. Plenty of data behind it.
101 ±18
Coding IQ 100.6, 95% interval 83–118. 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$10.001M
36.3
Poor value (36.3). Costs more per task than its benchmark results predict. Plenty of data behind it.
54 ±9
Typical (54 ±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.
144.9
Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API
Doubao Seed 2.0 Litedoubao-seed-2-0-lite-260215 · 260215
≈ 129
From 4 benchmarks. Little data behind it, so treat it as a rough estimate.
Not enough data yet$0.0904cache read $0.0181$0.5425256KNot enough data yetNot measuredNot measured
Doubao Seed 2.0 Litedoubao-seed-2-0-lite-260428 · 260428
≈ 129
From 4 benchmarks. Little data behind it, so treat it as a rough estimate.
Not enough data yet$0.0904cache read $0.0181$0.5425256KNot enough data yetNot measuredNot measured
Grok Build 0.1
🇺🇸
Maker headquartered in US, according to the company's own website. The citation is on the model page.
128.7
From 32 benchmarks. A fair amount of data behind it.
≈ 110 ±23
Coding IQ 109.5, 95% interval 87–133. Little data behind it, so treat it as a rough estimate. Uses the model's best published results. It has its own scale, separate from llminfo IQ.
$1.00cache read $0.20$2.00256K
57.8
Good value (57.8). Cheaper per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate.
Not measuredNot measured
DeepSeek V3.2 Speciale
🇨🇳
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.
azure-deepseek-v3.2-speciale
128.6
From 33 benchmarks. Plenty of data behind it.
Not enough data yet$0.58$1.68Not published by the provider
73.7
Good value (73.7). Cheaper per task than its benchmark results predict. Plenty of data behind it.
Not measuredNot measured
128.5
From 60 benchmarks. Plenty of data behind it.
Not enough data yet$0.4648cache read $0.093$2.3238256KNot enough data yetNot measuredNot measured

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

How this data works