Model intelligence
678 AI models
Capability (Epoch Capabilities Index and individual benchmarks from Epoch AI), API list prices and context windows (OpenRouter), release dates and open-weight status. Benchmarks are shown individually - never merged into one invented score.
Epoch Capabilities Index by release date
Capability frontier
Source: Epoch AI (CC BY 4.0). ECI is Epoch's item-response model over many benchmarks; each model page shows its 90% confidence interval.
Blended $/1M tokens (3:1 input:output) vs ECI
Price vs capability
Only models with both an ECI score and an OpenRouter price. Normalisation: 75% input + 25% output price. Click a point to open the model.
| # | Model | ECI ↓ | Context | In $/M | Weights | Released |
|---|---|---|---|---|---|---|
| 101 | Gemma 4 26B A4B Google· reasoning | 141.8 | 262K | $0.076 | Open source | 2 Apr 2026 |
| 102 | Claude Sonnet 4 Anthropic· reasoning | 141.7 | 200K | $3.00 | Closed | 22 May 2025 |
| 103 | Gemini 2.5 Flash (May 2025) | 141.5 | — | — | Closed | 20 May 2025 |
| 104 | Mistral Medium 3.5 Mistral AI· reasoning | 141.4 | 262K | $1.50 | Open source | 28 Apr 2026 |
| 105 | DeepSeek-R1 (May 2025) | 141.3 | — | — | Open source | 28 May 2025 |
| 106 | Claude 3.7 Sonnet | 141.2 | — | — | Closed | 24 Feb 2025 |
| 107 | Gemini 2.5 Flash (Jun 2025) | 140.8 | — | — | Closed | 17 Jun 2025 |
| 108 | Grok-3 mini | 140.3 | — | — | Closed | 24 Jun 2025 |
| 109 | o3 Mini OpenAI· reasoning | 140.3 | 200K | $1.10 | Closed | 31 Jan 2025 |
| 110 | Kimi K2 (Jul 2025) | 140.1 | — | — | Open source | 12 Jul 2025 |
| 111 | Gemini 2.5 Flash (Apr 2025) | 140.0 | — | — | Closed | 17 Apr 2025 |
| 112 | gpt-oss-120b OpenAI· reasoning | 139.9 | 131K | $0.037 | Open source | 5 Aug 2025 |
| 113 | DeepSeek V3.1 DeepSeek· reasoning | 139.9 | 164K | $0.25 | Open source | 21 Aug 2025 |
| 114 | Qwen3-30B-A3B-Thinking (Jul 2025) | 139.6 | — | — | Open source | 30 Jul 2025 |
| 115 | Qwen3.5-9B Alibaba (Qwen)· reasoning | 139.4 | 262K | $0.10 | Open source | 24 Feb 2026 |
| 116 | GPT-5 Nano OpenAI· reasoning | 139.4 | 400K | $0.050 | Closed | 7 Aug 2025 |
| 117 | Qwen3 235B A22B Alibaba (Qwen)· reasoning | 139.4 | 131K | $0.46 | Open source | 28 Apr 2025 |
| 118 | R1 DeepSeek· reasoning | 139.0 | 64K | $0.70 | Open source | 20 Jan 2025 |
| 119 | Qwen3-235B-A22B-Instruct (Jul 2025) | 138.9 | — | — | Open source | 25 Jul 2025 |
| 120 | Qwen3 32B Alibaba (Qwen)· reasoning | 138.5 | 131K | $0.080 | Open source | 29 Apr 2025 |
| 121 | Grok 3 | 138.3 | — | — | Closed | 9 Apr 2025 |
| 122 | Qwen3 14B Alibaba (Qwen)· reasoning | 138.2 | 131K | $0.12 | Open source | 29 Apr 2025 |
| 123 | gpt-oss-20b OpenAI· reasoning | 137.8 | 131K | $0.018 | Open source | 5 Aug 2025 |
| 124 | QwQ-32B | 137.6 | — | — | Open source | 5 Mar 2025 |
| 125 | DeepSeek-R1-Distill-Qwen-32B | 137.4 | — | — | Open source | 20 Jan 2025 |
| 126 | Qwen3-30B-A3B-Instruct (Jul 2025) | 137.4 | — | — | Open source | 29 Jul 2025 |
| 127 | GPT-4.1 | 136.8 | 1.05M | $2.00 | Closed | 14 Apr 2025 |
| 128 | GPT-4.5 | 136.7 | — | — | Closed | 27 Feb 2025 |
| 129 | Qwen3 30B A3B Alibaba (Qwen)· reasoning | 136.2 | 131K | $0.13 | Open source | 29 Apr 2025 |
| 130 | Qwen3 8B Alibaba (Qwen)· reasoning | 136.2 | 131K | $0.12 | Open source | 28 Apr 2025 |
| 131 | DeepSeek-V3 (Mar 2025) | 135.9 | — | — | Open source | 24 Mar 2025 |
| 132 | o1-mini | 135.8 | — | — | Closed | 12 Sep 2024 |
| 133 | DeepSeek-R1-Distill-Qwen-14B | 135.4 | — | — | Open source | 20 Jan 2025 |
| 134 | Gemini 2.0 Flash Thinking (Jan 2025) | 135.4 | — | — | Closed | 21 Jan 2025 |
| 135 | Gemini 2.0 Pro | 135.1 | — | — | Closed | 5 Feb 2025 |
| 136 | GPT-4.1 Mini | 135.0 | 1.05M | $0.40 | Closed | 14 Apr 2025 |
| 137 | o1-preview | 134.8 | — | — | Closed | 12 Sep 2024 |
| 138 | Gemini 2.0 Flash (Dec 2024) | 134.7 | — | — | Closed | 11 Dec 2024 |
| 139 | Gemini 2.0 Flash (Feb 2025) | 134.7 | — | — | Closed | 5 Feb 2025 |
| 140 | Mistral Medium 3 | 134.1 | 131K | $0.40 | Closed | 7 May 2025 |
| 141 | Gemini 2.5 Flash-Lite (Jun 2025) | 133.9 | — | — | Closed | 17 Jun 2025 |
| 142 | Claude 3.5 Sonnet (October 2024) | 133.6 | — | — | Closed | 22 Oct 2024 |
| 143 | Magistral Small 1.0 | 133.2 | — | — | Open source | 10 Jun 2025 |
| 144 | Qwen2.5-Max | 132.5 | — | — | Closed | 25 Jan 2025 |
| 145 | DeepSeek V3 | 132.4 | 164K | $0.26 | Open source | 26 Dec 2024 |
| 146 | Llama 4 Maverick | 132.2 | 1.05M | $0.19 | Open source | 6 Apr 2025 |
| 147 | Mistral Small 3.2 | 131.7 | — | — | Open source | 20 Jun 2025 |
| 148 | Gemini 1.5 Pro (Sept 2024) | 131.7 | — | — | Closed | 24 Sep 2024 |
| 149 | Magistral Small 1.2 | 131.4 | — | — | Open source | 18 Sep 2025 |
| 150 | Grok-2 (Dec 2024) | 130.5 | — | — | Closed | 12 Dec 2024 |
| 151 | Phi 4 | 130.4 | 16K | $0.070 | Open source | 12 Dec 2024 |
| 152 | Gemma 3 27B | 130.0 | 131K | $0.080 | Open source | 12 Mar 2025 |
| 153 | Claude 3.5 Sonnet | 130.0 | — | — | Closed | 20 Jun 2024 |
| 154 | Llama 4 Scout | 129.6 | 1.31M | $0.10 | Open source | 5 Apr 2025 |
| 155 | GPT-4.1 Nano | 129.6 | 1.05M | $0.10 | Closed | 14 Apr 2025 |
| 156 | Gemini 1.5 Flash (Sep 2024) | 129.4 | — | — | Closed | 24 Sep 2024 |
| 157 | Qwen2.5-72B | 129.0 | — | — | Open source | 19 Sep 2024 |
| 158 | GPT-4o (May 2024) | 129.0 | — | — | Closed | 13 May 2024 |
| 159 | GPT-4o (Nov 2024) | 128.8 | — | — | Closed | 20 Nov 2024 |
| 160 | GPT-4o (Aug 2024) | 128.8 | — | — | Closed | 6 Aug 2024 |
| 161 | Llama 3.1-405B | 128.8 | — | — | Open source | 23 Jul 2024 |
| 162 | Mistral Large 2 (Nov 2024) | 128.5 | — | — | Open source | 18 Nov 2024 |
| 163 | Qwen2.5-32B | 128.5 | — | — | Open source | 17 Sep 2024 |
| 164 | Mistral Large 2 (Jul 2024) | 127.5 | — | — | Open source | 24 Jul 2024 |
| 165 | Mistral Small 3.1 | 127.5 | — | — | Open source | 17 Mar 2025 |
| 166 | Llama 3.3 70B | 127.3 | — | — | Open source | 6 Dec 2024 |
| 167 | GPT-4 Turbo (Apr 2024) | 127.3 | — | — | Closed | 9 Apr 2024 |
| 168 | Claude 3.5 Haiku | 127.2 | — | — | Closed | 22 Oct 2024 |
| 169 | Mistral Small 3 | 127.1 | 33K | $0.050 | Open source | 30 Jan 2025 |
| 170 | Claude 3 Opus | 126.9 | — | — | Closed | 29 Feb 2024 |
| 171 | Gemini 1.5 Pro (May 2024) | 126.9 | — | — | Closed | 14 May 2024 |
| 172 | GPT-4o-mini | 126.6 | 128K | $0.15 | Closed | 18 Jul 2024 |
| 173 | GPT-4 Turbo (Nov 2023) | 126.5 | — | — | Closed | 25 Jan 2024 |
| 174 | Llama 3.1-70B | 125.9 | — | — | Open source | 23 Jul 2024 |
| 175 | GPT-4 (Mar 2023) | 125.9 | — | — | Closed | 14 Mar 2023 |
| 176 | Llama 3.2 90B | 125.5 | — | — | Open source | 24 Sep 2024 |
| 177 | Qwen2-72B | 125.3 | — | — | Open source | 7 Jun 2024 |
| 178 | DeepSeek-V2 (MoE-236B, May 2024) | 124.8 | — | — | Open source | 7 May 2024 |
| 179 | Amazon Nova Pro | 123.8 | — | — | Closed | 3 Dec 2024 |
| 180 | Gemma 3 12B | 123.5 | 131K | $0.050 | Open source | 12 Mar 2025 |
| 181 | GPT-4 (Jun 2023) | 123.1 | — | — | Closed | 13 Jun 2023 |
| 182 | Llama 3-70B | 122.9 | — | — | Open source | 18 Apr 2024 |
| 183 | Gemini 1.5 Flash (May 2024) | 122.6 | — | — | Closed | 23 May 2024 |
| 184 | Gemma 2 27B | 122.1 | 8K | $0.65 | Open source | 24 Jun 2024 |
| 185 | Mixtral 8x22B | 122.0 | — | — | Open source | 17 Apr 2024 |
| 186 | Mistral Large | 122.0 | 128K | $2.00 | Closed | 26 Feb 2024 |
| 187 | phi-3-small 7.4B | 121.8 | — | — | Open source | 23 Apr 2024 |
| 188 | phi-3-medium 14B | 121.2 | — | — | Open source | 23 Apr 2024 |
| 189 | Claude 3 Sonnet | 120.7 | — | — | Closed | 29 Feb 2024 |
| 190 | Claude Instant | 120.2 | — | — | Closed | 9 Aug 2023 |
| 191 | Claude 2 | 120.1 | — | — | Closed | 11 Jul 2023 |
| 192 | Gemma 2 9B | 119.8 | — | — | Open source | 24 Jun 2024 |
| 193 | Qwen2.5-Coder-32B | 119.4 | — | — | Open source | 18 Sep 2024 |
| 194 | Command R+ | 119.3 | — | — | Open source | 30 Aug 2024 |
| 195 | Claude 2.1 | 119.2 | — | — | Closed | 21 Nov 2023 |
| 196 | Mistral Nemo | 118.6 | 131K | $0.019 | Open source | 18 Jul 2024 |
| 197 | GPT-3.5 Turbo (Nov 2023) | 118.5 | — | — | Closed | 6 Nov 2023 |
| 198 | Qwen2.5-7B | 118.4 | — | — | Open source | 19 Sep 2024 |
| 199 | Mixtral 8x7B | 118.4 | — | — | Open source | 11 Dec 2023 |
| 200 | Claude 3 Haiku | 118.3 | — | — | Closed | 7 Mar 2024 |
Showing 100 of 678. Benchmark cells are % (best reported setting). 3.8K benchmark results in total.← PrevNext →Export CSV