Listings
All listings: Amazon Web Services
Every listing for the models on page 4 of 9 of Amazon Web Services, including each model's other versions and routes. Back to Amazon Web Services
Sold by Amazon Web Services
53 listings on this page sold by Amazon Web Services.
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
- Epoch Capabilities Index
- Epoch Capabilities Index benchmark details
| Model | llminfo IQ | Coding IQ | Input $/1M | Output $/1M | Max Input Tokens | Value | Yap | Epoch Capabilities Index |
|---|---|---|---|---|---|---|---|---|
Claude Haiku 3.5 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. Newer version | 110.1From 238 benchmarks. Plenty of data behind it. | 79 ±8Coding IQ 79.4, 95% interval 72–87. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | $0.80cache read $0.08 | $4.00cache write $1.00 | —Not published by the provider | 19.4Poor value (19.4). Costs more per task than its benchmark results predict. Plenty of data behind it. | —Not measured | 124.8Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API |
Claude Haiku 3.5 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. Newer version | 110.1From 238 benchmarks. Plenty of data behind it. | 79 ±8Coding IQ 79.4, 95% interval 72–87. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | $0.80 | $4.00 | —Not published by the provider | 19.4Poor value (19.4). Costs more per task than its benchmark results predict. Plenty of data behind it. | —Not measured | 124.8Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API |
Claude Haiku 3.5 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. Newer version | 110.1From 238 benchmarks. Plenty of data behind it. | 79 ±8Coding IQ 79.4, 95% interval 72–87. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | $0.80cache read $0.08 | $4.00cache write $1.00 | 200K | 19.4Poor value (19.4). Costs more per task than its benchmark results predict. Plenty of data behind it. | —Not measured | 124.8Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API |
Claude Haiku 3.5 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. Newer version | 110.1From 238 benchmarks. Plenty of data behind it. | 79 ±8Coding IQ 79.4, 95% interval 72–87. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | $0.80cache read $0.08 | $4.00cache write $1.00 | 200K | 19.4Poor value (19.4). Costs more per task than its benchmark results predict. Plenty of data behind it. | —Not measured | 124.8Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API |
Claude Haiku 3.5 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. Newer version | 110.1From 238 benchmarks. Plenty of data behind it. | 79 ±8Coding IQ 79.4, 95% interval 72–87. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | $0.80cache read $0.08 | $4.00cache write $1.00 | 200K | 19.4Poor value (19.4). Costs more per task than its benchmark results predict. Plenty of data behind it. | —Not measured | 124.8Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API |
Claude Haiku 3.5 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. Newer version | 110.1From 238 benchmarks. Plenty of data behind it. | 79 ±8Coding IQ 79.4, 95% interval 72–87. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | $0.80 | $4.00 | —Not published by the provider | 19.4Poor value (19.4). Costs more per task than its benchmark results predict. Plenty of data behind it. | —Not measured | 124.8Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API |
Magistral Small 2509 🇫🇷Maker headquartered in FR, 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. | 109.2From 42 benchmarks. Plenty of data behind it. | —Not enough data yet | $0.50batch $0.25 | $1.50batch $0.75 | 128K | 27.6Poor value (27.6). Costs more per task than its benchmark results predict. Plenty of data behind it. | —Not measured | —Not measured |
Gemma 3 27B IT 🇺🇸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. | 108.5From 365 benchmarks. Plenty of data behind it. | 73 ±8Coding IQ 72.7, 95% interval 65–81. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | $0.23batch $0.12 | $0.38batch $0.19 | 128K | 46.2Poor value (46.2). Costs more per task than its benchmark results predict. A fair amount of data behind it. | —Not measured | 127.6Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API |
Llama 3.3 70B Instruct 🇺🇸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. | 108.1From 278 benchmarks. Plenty of data behind it. | 77 ±9Coding IQ 77.1, 95% interval 68–86. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | $0.72batch $0.36 | $0.72batch $0.36 | 128K | 31.6Poor value (31.6). Costs more per task than its benchmark results predict. A fair amount of data behind it. | —Not measured | 126.2Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API |
Mistral Large 🇫🇷Maker headquartered in FR, according to the company's own website. The citation is on the model page. | 108.0From 293 benchmarks. Plenty of data behind it. | 72 ±9Coding IQ 72.2, 95% interval 63–81. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | $4.00 | $12.00 | 32K | 22.0Poor value (22.0). Costs more per task than its benchmark results predict. Plenty of data behind it. | —Not measured | 120Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API |
Mistral Large 🇫🇷Maker headquartered in FR, according to the company's own website. The citation is on the model page. | 108.0From 293 benchmarks. Plenty of data behind it. | 72 ±9Coding IQ 72.2, 95% interval 63–81. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | $2.00 | $6.00 | 128K | 22.0Poor value (22.0). Costs more per task than its benchmark results predict. Plenty of data behind it. | —Not measured | 120Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API |
Gemma 3 12B IT 🇺🇸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. Newer version | 107.6From 331 benchmarks. Plenty of data behind it. | 68 ±19Coding IQ 68.1, 95% interval 49–87. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | $0.09batch $0.05 | $0.29batch $0.15 | 128K | 41.7Poor value (41.7). Costs more per task than its benchmark results predict. A fair amount of data behind it. | —Not measured | —Not measured |
Llama 3.1 70B Instruct 🇺🇸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. Newer version | 106.8From 168 benchmarks. Plenty of data behind it. | ≈ 80 ±11Coding IQ 80.3, 95% interval 69–91. 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. | $0.72batch $0.36 | $0.72batch $0.36 | 128K | 41.0Poor value (41.0). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured |
Ministral 3 3b Instruct 🇫🇷Maker headquartered in FR, 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. | 106.4From 8 benchmarks. A fair amount of data behind it. | —Not enough data yet | $0.10batch $0.05 | $0.10batch $0.05 | 128K | —Not enough data yet | —Not measured | —Not measured |
Nova Pro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 106.4From 119 benchmarks. Plenty of data behind it. | —Not enough data yet | $0.80cache read $0.20batch $0.40 | $3.20batch $1.60 | 300K | 36.8Poor value (36.8). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured |
Nova Pro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 106.4From 119 benchmarks. Plenty of data behind it. | —Not enough data yet | — | — | —Not published by the provider | 36.8Poor value (36.8). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured |
Nova Pro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 106.4From 119 benchmarks. Plenty of data behind it. | —Not enough data yet | — | — | —Not published by the provider | 36.8Poor value (36.8). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured |
Nova Pro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 106.4From 119 benchmarks. Plenty of data behind it. | —Not enough data yet | $0.80cache read $0.2175 | $3.20cache write $0.92 | 300K | 36.8Poor value (36.8). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured |
Nova Pro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 106.4From 119 benchmarks. Plenty of data behind it. | —Not enough data yet | $0.80cache read $0.20batch $0.40 | $3.20batch $1.60 | 300K | 36.8Poor value (36.8). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured |
Nova Pro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 106.4From 119 benchmarks. Plenty of data behind it. | —Not enough data yet | $0.87cache read $0.2175 | $3.48 | 300K | 36.8Poor value (36.8). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured |
Nova Pro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 106.4From 119 benchmarks. Plenty of data behind it. | —Not enough data yet | $0.80 | $3.20 | 300K | 36.8Poor value (36.8). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured |
Nova Pro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 106.4From 119 benchmarks. Plenty of data behind it. | —Not enough data yet | $0.80 | $3.20 | —Not published by the provider | 36.8Poor value (36.8). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured |
Llama 4 Scout 🇺🇸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. | 106.2From 347 benchmarks. Plenty of data behind it. | 69 ±8Coding IQ 69.3, 95% interval 61–78. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | $0.17batch $0.085 | $0.66batch $0.33 | 10M | 30.3Poor value (30.3). Costs more per task than its benchmark results predict. Plenty of data behind it. | —Not measured | 127.8Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API |
Gemma 4 E2B IT 🇺🇸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. | ≈ 106From 9 benchmarks. Little data behind it, so treat it as a rough estimate. | —Not enough data yet | $0.04batch $0.02 | $0.08batch $0.04 | 128K | —Not enough data yet | —Not measured | —Not measured |
Nova Lite 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 104.7From 107 benchmarks. Plenty of data behind it. | ≈ 82 ±10Coding IQ 82.3, 95% interval 72–93. 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. | $0.06cache read $0.015batch $0.03 | $0.24batch $0.12 | 300K | 41.5Poor value (41.5). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | 38 ±9Terse (38 ±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. | —Not measured |
Nova Lite 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 104.7From 107 benchmarks. Plenty of data behind it. | ≈ 82 ±10Coding IQ 82.3, 95% interval 72–93. 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. | — | — | —Not published by the provider | 41.5Poor value (41.5). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | 38 ±9Terse (38 ±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. | —Not measured |
Nova Lite 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 104.7From 107 benchmarks. Plenty of data behind it. | ≈ 82 ±10Coding IQ 82.3, 95% interval 72–93. 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. | — | — | —Not published by the provider | 41.5Poor value (41.5). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | 38 ±9Terse (38 ±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. | —Not measured |
Nova Lite 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 104.7From 107 benchmarks. Plenty of data behind it. | ≈ 82 ±10Coding IQ 82.3, 95% interval 72–93. 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. | $0.06cache read $0.0163 | $0.24cache write $0.069 | 300K | 41.5Poor value (41.5). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | 38 ±9Terse (38 ±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. | —Not measured |
Nova Lite 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 104.7From 107 benchmarks. Plenty of data behind it. | ≈ 82 ±10Coding IQ 82.3, 95% interval 72–93. 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. | $0.06cache read $0.015batch $0.03 | $0.24batch $0.12 | 300K | 41.5Poor value (41.5). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | 38 ±9Terse (38 ±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. | —Not measured |
Nova Lite 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 104.7From 107 benchmarks. Plenty of data behind it. | ≈ 82 ±10Coding IQ 82.3, 95% interval 72–93. 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. | $0.075cache read $0.0188 | $0.30 | 300K | 41.5Poor value (41.5). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | 38 ±9Terse (38 ±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. | —Not measured |
Nova Lite 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 104.7From 107 benchmarks. Plenty of data behind it. | ≈ 82 ±10Coding IQ 82.3, 95% interval 72–93. 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. | $0.064cache read $0.016 | $0.256 | 300K | 41.5Poor value (41.5). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | 38 ±9Terse (38 ±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. | —Not measured |
Nova Lite 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 104.7From 107 benchmarks. Plenty of data behind it. | ≈ 82 ±10Coding IQ 82.3, 95% interval 72–93. 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. | $0.06 | $0.24 | 300K | 41.5Poor value (41.5). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | 38 ±9Terse (38 ±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. | —Not measured |
Claude 3 Sonnet 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. Newer version | 103.7From 286 benchmarks. Plenty of data behind it. | ≈ 76 ±10Coding IQ 75.7, 95% interval 66–86. 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. | $3.00cache read $0.30 | $15.00cache write $3.75 | —Not published by the provider | 38.2Poor value (38.2). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | 119.7Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API |
Claude 3 Sonnet 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. Newer version | 103.7From 286 benchmarks. Plenty of data behind it. | ≈ 76 ±10Coding IQ 75.7, 95% interval 66–86. 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. | $3.00 | $15.00 | —Not published by the provider | 38.2Poor value (38.2). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | 119.7Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API |
Claude 3 Sonnet 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. Newer version | 103.7From 286 benchmarks. Plenty of data behind it. | ≈ 76 ±10Coding IQ 75.7, 95% interval 66–86. 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. | $3.00 | $15.00 | —Not published by the provider | 38.2Poor value (38.2). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | 119.7Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API |
Claude 2 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. Newer version | 103.6From 32 benchmarks. Plenty of data behind it. | —Not enough data yet | $8.00 | $24.00 | —Not published by the provider | —Not enough data yet | —Not measured | 117.2Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API |
Nova Micro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 103.3From 102 benchmarks. Plenty of data behind it. | —Not enough data yet | $0.035cache read $0.0088batch $0.0175 | $0.14batch $0.07 | 128K | 47.4Poor value (47.4). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured |
Nova Micro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 103.3From 102 benchmarks. Plenty of data behind it. | —Not enough data yet | — | — | —Not published by the provider | 47.4Poor value (47.4). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured |
Nova Micro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 103.3From 102 benchmarks. Plenty of data behind it. | —Not enough data yet | — | — | —Not published by the provider | 47.4Poor value (47.4). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured |
Nova Micro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 103.3From 102 benchmarks. Plenty of data behind it. | —Not enough data yet | $0.035cache read $0.0095 | $0.14cache write $0.04 | 128K | 47.4Poor value (47.4). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured |
Nova Micro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 103.3From 102 benchmarks. Plenty of data behind it. | —Not enough data yet | $0.035cache read $0.0088batch $0.0175 | $0.14batch $0.07 | 128K | 47.4Poor value (47.4). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured |
Nova Micro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 103.3From 102 benchmarks. Plenty of data behind it. | —Not enough data yet | $0.037cache read $0.0093 | $0.148cache write $0.037 | 128K | 47.4Poor value (47.4). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured |
Nova Micro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 103.3From 102 benchmarks. Plenty of data behind it. | —Not enough data yet | $0.035 | $0.14 | 128K | 47.4Poor value (47.4). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured |
Claude 3 Haiku 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. Newer version | 102.7From 315 benchmarks. Plenty of data behind it. | 76 ±10Coding IQ 75.9, 95% interval 66–86. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | $0.25 | $1.25 | —Not published by the provider | 33.4Poor value (33.4). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | 117.8Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API |
Claude 3 Haiku 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. Newer version | 102.7From 315 benchmarks. Plenty of data behind it. | 76 ±10Coding IQ 75.9, 95% interval 66–86. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | $0.25 | $1.25 | —Not published by the provider | 33.4Poor value (33.4). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | 117.8Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API |
Claude 3 Haiku 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. Newer version | 102.7From 315 benchmarks. Plenty of data behind it. | 76 ±10Coding IQ 75.9, 95% interval 66–86. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | $0.30 | $1.50 | 196K | 33.4Poor value (33.4). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | 117.8Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API |
Claude 3 Haiku 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. Newer version | 102.7From 315 benchmarks. Plenty of data behind it. | 76 ±10Coding IQ 75.9, 95% interval 66–86. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | $0.25 | $1.25 | —Not published by the provider | 33.4Poor value (33.4). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | 117.8Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API |
Claude 2.1 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 101.5From 85 benchmarks. A fair amount of data behind it. | —Not enough data yet | $8.00 | $24.00 | 100K | 37.8Poor value (37.8). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured |
Llama 3.2 11B Instruct 🇺🇸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. | 100.4From 10 benchmarks. A fair amount of data behind it. | —Not enough data yet | $0.16 | $0.16 | 128K | —Not enough data yet | —Not measured | —Not measured |
Llama 3.2 11B Instruct 🇺🇸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. | 100.4From 10 benchmarks. A fair amount of data behind it. | —Not enough data yet | $0.16 | $0.16 | 128K | —Not enough data yet | —Not measured | —Not measured |
Llama 2 70B Chat 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 98.8From 50 benchmarks. A fair amount of data behind it. | —Not enough data yet | $1.95 | $2.56 | 4K | —Not enough data yet | —Not measured | —Not measured |
Command R+ 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 97.3From 220 benchmarks. Plenty of data behind it. | ≈ 76 ±11Coding IQ 75.7, 95% interval 65–87. 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. | $0.50 | $1.50 | —Not published by the provider | —Not enough data yet | —Not measured | —Not measured |
Command R Plus 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 96.9From 196 benchmarks. Plenty of data behind it. | —Not enough data yet | $3.00 | $15.00 | —Not published by the provider | 21.9Poor value (21.9). Costs more per task than its benchmark results predict. A fair amount of data behind it. | —Not measured | —Not measured |
Run by Amazon Web Services
28 listings on this page where Amazon Web Services runs the model. Another provider sells these listings.
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
- Epoch Capabilities Index
- Epoch Capabilities Index benchmark details
| Model | Provider | llminfo IQ | Coding IQ | Input $/1M | Output $/1M | Max Input Tokens | Value | Yap | Epoch Capabilities Index |
|---|---|---|---|---|---|---|---|---|---|
Nemotron 3 Nano 30B A3B 🇺🇸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. | 109.7From 137 benchmarks. Plenty of data behind it. | 91 ±14Coding IQ 91.4, 95% interval 77–106. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | — | — | 256K | 36.7Poor value (36.7). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured | |
Gemma 3 27B IT 🇺🇸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. | 108.5From 365 benchmarks. Plenty of data behind it. | 73 ±8Coding IQ 72.7, 95% interval 65–81. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | $0.23 | $0.38 | 128K | 46.2Poor value (46.2). Costs more per task than its benchmark results predict. A fair amount of data behind it. | —Not measured | 127.6Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API | |
Gemma 3 27B IT 🇺🇸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. | 108.5From 365 benchmarks. Plenty of data behind it. | 73 ±8Coding IQ 72.7, 95% interval 65–81. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | $0.23 | $0.38 | 128K | 46.2Poor value (46.2). Costs more per task than its benchmark results predict. A fair amount of data behind it. | —Not measured | 127.6Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API | |
Llama 3.3 70B Instruct 🇺🇸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. | 108.1From 278 benchmarks. Plenty of data behind it. | 77 ±9Coding IQ 77.1, 95% interval 68–86. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | $0.72 | $0.72 | 131K | 31.6Poor value (31.6). Costs more per task than its benchmark results predict. A fair amount of data behind it. | —Not measured | 126.2Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API | |
Llama 3.3 70B Instruct 🇺🇸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. | 108.1From 278 benchmarks. Plenty of data behind it. | 77 ±9Coding IQ 77.1, 95% interval 68–86. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | $0.72 | $0.72 | 131K | 31.6Poor value (31.6). Costs more per task than its benchmark results predict. A fair amount of data behind it. | —Not measured | 126.2Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API | |
Mistral Large 🇫🇷Maker headquartered in FR, according to the company's own website. The citation is on the model page. | 108.0From 293 benchmarks. Plenty of data behind it. | 72 ±9Coding IQ 72.2, 95% interval 63–81. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | — | — | 32K | 22.0Poor value (22.0). Costs more per task than its benchmark results predict. Plenty of data behind it. | —Not measured | 120Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API | |
Gemma 3 12B IT 🇺🇸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. Newer version | 107.6From 331 benchmarks. Plenty of data behind it. | 68 ±19Coding IQ 68.1, 95% interval 49–87. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | $0.09 | $0.29 | 128K | 41.7Poor value (41.7). Costs more per task than its benchmark results predict. A fair amount of data behind it. | —Not measured | —Not measured | |
Gemma 3 12B IT 🇺🇸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. Newer version | 107.6From 331 benchmarks. Plenty of data behind it. | 68 ±19Coding IQ 68.1, 95% interval 49–87. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | $0.09 | $0.29 | 128K | 41.7Poor value (41.7). Costs more per task than its benchmark results predict. A fair amount of data behind it. | —Not measured | —Not measured | |
Ministral 8B 🇫🇷Maker headquartered in FR, 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. | 107.6From 73 benchmarks. Plenty of data behind it. | —Not enough data yet | — | — | 256K | 54.6Good value (54.6). Cheaper per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured | |
Llama 3.1 70B Instruct 🇺🇸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. Newer version | 106.8From 168 benchmarks. Plenty of data behind it. | ≈ 80 ±11Coding IQ 80.3, 95% interval 69–91. 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. | $0.72 | $0.72 | 131K | 41.0Poor value (41.0). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured | |
Llama 3.1 70B Instruct 🇺🇸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. Newer version | 106.8From 168 benchmarks. Plenty of data behind it. | ≈ 80 ±11Coding IQ 80.3, 95% interval 69–91. 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. | $0.72 | $0.72 | —Not published by the provider | 41.0Poor value (41.0). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured | |
Llama 3.1 70B Instruct 🇺🇸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. Newer version | 106.8From 168 benchmarks. Plenty of data behind it. | ≈ 80 ±11Coding IQ 80.3, 95% interval 69–91. 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. | $0.72 | $0.72 | 131K | 41.0Poor value (41.0). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured | |
Nova Pro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 106.4From 119 benchmarks. Plenty of data behind it. | —Not enough data yet | — | — | 300K | 36.8Poor value (36.8). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured | |
Nova Pro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 106.4From 119 benchmarks. Plenty of data behind it. | —Not enough data yet | $0.80cache read $0.2175 | $3.20 | 300K | 36.8Poor value (36.8). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured | |
Nova Pro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 106.4From 119 benchmarks. Plenty of data behind it. | —Not enough data yet | $0.80cache read $0.2175 | $3.20 | 300K | 36.8Poor value (36.8). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured | |
Nova Pro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 106.4From 119 benchmarks. Plenty of data behind it. | —Not enough data yet | $0.80 | $3.20 | 300K | 36.8Poor value (36.8). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured | |
Nova Pro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 106.4From 119 benchmarks. Plenty of data behind it. | —Not enough data yet | $0.80 | $3.20 | 300K | 36.8Poor value (36.8). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured | |
Llama 4 Scout 🇺🇸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. | 106.2From 347 benchmarks. Plenty of data behind it. | 69 ±8Coding IQ 69.3, 95% interval 61–78. A fair amount of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ. | $0.17 | $0.66 | —Not published by the provider | 30.3Poor value (30.3). Costs more per task than its benchmark results predict. Plenty of data behind it. | —Not measured | 127.8Epoch Capabilities IndexSource: Epoch AI Benchmarking Hub API | |
Nova Lite 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 104.7From 107 benchmarks. Plenty of data behind it. | ≈ 82 ±10Coding IQ 82.3, 95% interval 72–93. 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. | — | — | 300K | 41.5Poor value (41.5). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | 38 ±9Terse (38 ±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. | —Not measured | |
Nova Lite 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 104.7From 107 benchmarks. Plenty of data behind it. | ≈ 82 ±10Coding IQ 82.3, 95% interval 72–93. 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. | $0.06cache read $0.0163 | $0.24 | 300K | 41.5Poor value (41.5). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | 38 ±9Terse (38 ±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. | —Not measured | |
Nova Lite 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 104.7From 107 benchmarks. Plenty of data behind it. | ≈ 82 ±10Coding IQ 82.3, 95% interval 72–93. 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. | $0.06cache read $0.0163 | $0.24 | 300K | 41.5Poor value (41.5). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | 38 ±9Terse (38 ±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. | —Not measured | |
Nova Lite 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 104.7From 107 benchmarks. Plenty of data behind it. | ≈ 82 ±10Coding IQ 82.3, 95% interval 72–93. 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. | $0.06 | $0.24 | 300K | 41.5Poor value (41.5). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | 38 ±9Terse (38 ±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. | —Not measured | |
Nova Lite 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 104.7From 107 benchmarks. Plenty of data behind it. | ≈ 82 ±10Coding IQ 82.3, 95% interval 72–93. 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. | $0.06 | $0.24 | 300K | 41.5Poor value (41.5). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | 38 ±9Terse (38 ±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. | —Not measured | |
Nova Micro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 103.3From 102 benchmarks. Plenty of data behind it. | —Not enough data yet | — | — | 128K | 47.4Poor value (47.4). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured | |
Nova Micro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 103.3From 102 benchmarks. Plenty of data behind it. | —Not enough data yet | $0.035cache read $0.011 | $0.14 | 128K | 47.4Poor value (47.4). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured | |
Nova Micro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 103.3From 102 benchmarks. Plenty of data behind it. | —Not enough data yet | $0.035cache read $0.011 | $0.14 | 128K | 47.4Poor value (47.4). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured | |
Nova Micro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 103.3From 102 benchmarks. Plenty of data behind it. | —Not enough data yet | $0.035 | $0.14 | 128K | 47.4Poor value (47.4). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured | |
Nova Micro 🇺🇸Maker headquartered in US, according to the company's own website. The citation is on the model page. | 103.3From 102 benchmarks. Plenty of data behind it. | —Not enough data yet | $0.035 | $0.14 | 128K | 47.4Poor value (47.4). Costs more per task than its benchmark results predict. Little data behind it, so treat it as a rough estimate. | —Not measured | —Not measured |
Data credits: Epoch AI Benchmarking Hub, CC BY 4.0 (licence)