Knowledge & reasoning
ARC-AGI-3 leaderboard
As of 2026-10-07, the highest measured score on ARC-AGI-3 is 62.7% by GPT-6 Astra. 222 more models have estimated scores, calibrated from the benchmarks they were measured on.
Measured scores: benchlm.ai.
| # | Model | Score | Source |
|---|---|---|---|
| 1 | Claude Fable 5.1 | 73.5% | estimated ± 3.4 pp, medium confidence |
| 2 | Claude Opus 5.5 | 73.3% | estimated ± 3.4 pp, medium confidence |
| 3 | Claude Fable 5 | 73.1% | estimated ± 3.4 pp, medium confidence |
| 4 | GPT-6 Astra | 62.7% | measured |
| 5 | GPT-6.1 Sol | 52.7% | measured |
| 6 | Sakana Fugu | 34.2% | estimated ± 2.1 pp, high confidence |
| 7 | Sakana Fugu-Ultra | 34.2% | estimated ± 2.1 pp, high confidence |
| 8 | Claude Opus 5 | 30.2% | measured |
| 9 | Gemini 3.8 Flash | 10.4% | measured |
| 10 | GPT-5.6 Sol | 7.8% | measured |
| 11 | GPT-5.4 Pro | 5.6% | estimated ± 11.5 pp, low confidence |
| 12 | GPT-6 Sol | 4.6% | measured |
| 13 | Claude Mythos 5 | 2.8% | estimated ± 2.1 pp, high confidence |
| 14 | Grok 4.6 | 2.1% | measured |
| 15 | Gemini 3 Pro | 1.8% | estimated ± 3.4 pp, high confidence |
| 16 | Gemini 3.7 Flash | 1.7% | estimated ± 3.4 pp, high confidence |
| 17 | Claude Sonnet 5.5 | 1.7% | estimated ± 3.4 pp, high confidence |
| 18 | GPT-5.3 Codex | 1.7% | estimated ± 3.4 pp, high confidence |
| 19 | Muse Spark 1.1 | 1.7% | estimated ± 3.4 pp, high confidence |
| 20 | A.X K2 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 21 | Apodex 1.1 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 22 | Apodex 1.1 Mini | 1.7% | estimated ± 3.4 pp, medium confidence |
| 23 | Celeris-1 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 24 | Claude 3 Haiku | 1.7% | estimated ± 3.4 pp, medium confidence |
| 25 | Claude 4 Sonnet | 1.7% | estimated ± 3.4 pp, medium confidence |
| 26 | Claude Opus 4.5 Thinking | 1.7% | estimated ± 3.4 pp, high confidence |
| 27 | Claude Opus 4.6 (Adaptive) | 1.7% | estimated ± 3.4 pp, high confidence |
| 28 | Claude Opus 4.7 | 1.7% | estimated ± 3.4 pp, high confidence |
| 29 | Claude Sonnet 5 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 30 | Command A+ | 1.7% | estimated ± 3.4 pp, medium confidence |
| 31 | DeepSeek-R1 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 32 | DeepSeek V3 0324 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 33 | DeepSeek V3.1 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 34 | DeepSeek V3.1 (Reasoning) | 1.7% | estimated ± 3.4 pp, medium confidence |
| 35 | DeepSeek V3.2 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 36 | Exaone 4.0 1.2B | 1.7% | estimated ± 3.4 pp, medium confidence |
| 37 | Exaone 4.0 32B | 1.7% | estimated ± 3.4 pp, medium confidence |
| 38 | Gemini 2.5 Flash | 1.7% | estimated ± 3.4 pp, medium confidence |
| 39 | Gemini 3.5 Flash-Lite | 1.7% | estimated ± 3.4 pp, medium confidence |
| 40 | Gemini 3.6 Flash | 1.7% | estimated ± 3.4 pp, high confidence |
| 41 | Gemini 3 Flash | 1.7% | estimated ± 3.4 pp, high confidence |
| 42 | Gemini 4 Argon | 1.7% | estimated ± 3.4 pp, high confidence |
| 43 | Gemma 3 27B | 1.7% | estimated ± 3.4 pp, medium confidence |
| 44 | Gemma 4 26B A4B | 1.7% | estimated ± 3.4 pp, medium confidence |
| 45 | GLM-4.5-Air | 1.7% | estimated ± 3.4 pp, medium confidence |
| 46 | GLM-4.6 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 47 | GLM-5.1 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 48 | GLM-5.3 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 49 | GLM-5-Turbo | 1.7% | estimated ± 3.4 pp, medium confidence |
| 50 | GLM-5V-Turbo | 1.7% | estimated ± 3.4 pp, medium confidence |
| 51 | GPT-4o | 1.7% | estimated ± 3.4 pp, medium confidence |
| 52 | GPT-5.1 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 53 | GPT-5.1-Codex | 1.7% | estimated ± 3.4 pp, medium confidence |
| 54 | GPT-5.1-Codex-Max | 1.7% | estimated ± 3.4 pp, medium confidence |
| 55 | GPT-5.2-Codex | 1.7% | estimated ± 3.4 pp, medium confidence |
| 56 | GPT-5 (high) | 1.7% | estimated ± 3.4 pp, medium confidence |
| 57 | GPT-5 (medium) | 1.7% | estimated ± 3.4 pp, medium confidence |
| 58 | GPT-OSS 120B | 1.7% | estimated ± 3.4 pp, medium confidence |
| 59 | GPT-OSS 20B | 1.7% | estimated ± 3.4 pp, medium confidence |
| 60 | Granite-4.0-350M | 1.7% | estimated ± 3.4 pp, medium confidence |
| 61 | Granite-4.0-H-1B | 1.7% | estimated ± 3.4 pp, medium confidence |
| 62 | Granite-4.0-H-350M | 1.7% | estimated ± 3.4 pp, medium confidence |
| 63 | Grok 4 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 64 | Grok 4.1 Fast | 1.7% | estimated ± 3.4 pp, medium confidence |
| 65 | Grok 4.1 Fast (Reasoning) | 1.7% | estimated ± 3.4 pp, medium confidence |
| 66 | Grok 4.7 | 1.7% | estimated ± 3.4 pp, high confidence |
| 67 | Grok 4 Fast (Reasoning) | 1.7% | estimated ± 3.4 pp, medium confidence |
| 68 | Grok Code Fast 1 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 69 | Hy3 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 70 | K-Exaone | 1.7% | estimated ± 3.4 pp, medium confidence |
| 71 | K-EXAONE 2.0 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 72 | Kimi K2 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 73 | Kimi K2.7 Code | 1.7% | estimated ± 3.4 pp, medium confidence |
| 74 | LFM2.5-2.6B | 1.7% | estimated ± 3.4 pp, medium confidence |
| 75 | LFM2.5-8B-A1B | 1.7% | estimated ± 3.4 pp, medium confidence |
| 76 | LFM2.5-VL-1.6B-Extract | 1.7% | estimated ± 3.4 pp, medium confidence |
| 77 | Ling 3.0 Flash VL | 1.7% | estimated ± 3.4 pp, medium confidence |
| 78 | Ling 3.0 Tiny | 1.7% | estimated ± 3.4 pp, medium confidence |
| 79 | Ling 3.1 Flash | 1.7% | estimated ± 3.4 pp, medium confidence |
| 80 | Llama 3.1 405B | 1.7% | estimated ± 3.4 pp, medium confidence |
| 81 | Llama 4 Maverick | 1.7% | estimated ± 3.4 pp, medium confidence |
| 82 | Llama 4 Scout | 1.7% | estimated ± 3.4 pp, medium confidence |
| 83 | Mercury 2.5 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 84 | MiMo-V2.5-Pro | 1.7% | estimated ± 3.4 pp, medium confidence |
| 85 | MiMo-V2.6-Flash | 1.7% | estimated ± 3.4 pp, medium confidence |
| 86 | MiMo-V2.6-Pro | 1.7% | estimated ± 3.4 pp, medium confidence |
| 87 | MiMo-V2-Omni | 1.7% | estimated ± 3.4 pp, medium confidence |
| 88 | MiMo-V2-Pro | 1.7% | estimated ± 3.4 pp, medium confidence |
| 89 | MiniCPM5-2B | 1.7% | estimated ± 3.4 pp, medium confidence |
| 90 | MiniMax M2.7 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 91 | MiniMax M3 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 92 | Mistral Large 2 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 93 | Mistral Large 3 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 94 | Mistral Large 4 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 95 | Mistral Medium 3 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 96 | Mistral Medium 3.5 128B | 1.7% | estimated ± 3.4 pp, medium confidence |
| 97 | Mistral Small 4 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 98 | Mistral Small 4 (Reasoning) | 1.7% | estimated ± 3.4 pp, medium confidence |
| 99 | Muse Glimmer 30B | 1.7% | estimated ± 3.4 pp, medium confidence |
| 100 | Muse Spark | 1.7% | estimated ± 3.4 pp, high confidence |
| 101 | Muse Spark 1.2 | 1.7% | estimated ± 3.4 pp, high confidence |
| 102 | Muse Spark 1.3 | 1.7% | estimated ± 3.4 pp, high confidence |
| 103 | Nemotron 3 Nano 30B | 1.7% | estimated ± 3.4 pp, medium confidence |
| 104 | Nemotron 3 Super 100B | 1.7% | estimated ± 3.4 pp, medium confidence |
| 105 | Nemotron Ultra 253B | 1.7% | estimated ± 3.4 pp, medium confidence |
| 106 | North Mini Code | 1.7% | estimated ± 3.4 pp, medium confidence |
| 107 | Nova Pro | 1.7% | estimated ± 3.4 pp, medium confidence |
| 108 | o3 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 109 | Phi-4 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 110 | Quasar 438B | 1.7% | estimated ± 3.4 pp, medium confidence |
| 111 | Qwen3.5 397B (Reasoning) | 1.7% | estimated ± 3.4 pp, medium confidence |
| 112 | Qwen 3.6 Max (preview) | 1.7% | estimated ± 3.4 pp, medium confidence |
| 113 | Qwen3.8 Max Preview | 1.7% | estimated ± 3.4 pp, medium confidence |
| 114 | Qwen3 Max | 1.7% | estimated ± 3.4 pp, medium confidence |
| 115 | Qwen3-Omni-30B-A3B-Instruct | 1.7% | estimated ± 3.4 pp, medium confidence |
| 116 | Qwen3-Omni-30B-A3B-Thinking | 1.7% | estimated ± 3.4 pp, medium confidence |
| 117 | Sarvam 105B | 1.7% | estimated ± 3.4 pp, medium confidence |
| 118 | Sarvam 30B | 1.7% | estimated ± 3.4 pp, medium confidence |
| 119 | Solar Pro 2 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 120 | Solar Pro 3 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 121 | Solar Pro 4 | 1.7% | estimated ± 3.4 pp, medium confidence |
| 122 | Step 3.7 Flash | 1.7% | estimated ± 3.4 pp, medium confidence |
| 123 | Step 5 Preview | 1.7% | estimated ± 3.4 pp, medium confidence |
| 124 | Trinity-Large-Preview | 1.7% | estimated ± 3.4 pp, medium confidence |
| 125 | Trinity-Large-Thinking | 1.7% | estimated ± 3.4 pp, medium confidence |
| 126 | Ultravox v0.6 Llama 3.3 70B | 1.7% | estimated ± 3.4 pp, medium confidence |
| 127 | Claude Haiku 5.5 | 1.6% | estimated ± 10.7 pp, low confidence |
| 128 | Claude 3 Opus | 1.6% | estimated ± 10.7 pp, low confidence |
| 129 | Claude 4.1 Opus Thinking | 1.6% | estimated ± 10.7 pp, low confidence |
| 130 | DeepSeek R1 Distill Qwen 32B | 1.6% | estimated ± 10.7 pp, low confidence |
| 131 | Gemini 1.0 Pro | 1.6% | estimated ± 10.7 pp, low confidence |
| 132 | Gemini 1.5 Pro | 1.6% | estimated ± 10.7 pp, low confidence |
| 133 | GPT-4 Turbo | 1.6% | estimated ± 10.7 pp, low confidence |
| 134 | GPT-4o mini | 1.6% | estimated ± 10.7 pp, low confidence |
| 135 | Phi-4 Multimodal Instruct | 1.6% | estimated ± 10.7 pp, low confidence |
| 136 | Qwen2.5 Coder 32B Instruct | 1.6% | estimated ± 10.7 pp, low confidence |
| 137 | Claude Opus 4.8 | 1.5% | measured |
| 138 | Kimi K3 | 0.9% | estimated ± 2.1 pp, high confidence |
| 139 | GPT-5.6 Terra | 0.8% | measured |
| 140 | GPT-5.5 Pro | 0.6% | estimated ± 11.5 pp, low confidence |
| 141 | Pareto 26.9 | 0.5% | estimated ± 11.6 pp, low confidence |
| 142 | GPT-5.5 | 0.4% | measured |
| 143 | Gemini 3.1 Pro | 0.4% | measured |
| 144 | Grok 4.5 | 0.3% | measured |
| 145 | Agents-A1 | 0.3% | estimated ± 11.5 pp, low confidence |
| 146 | dots3-note Preview | 0.3% | estimated ± 11.5 pp, low confidence |
| 147 | Ornith-1.5-397B | 0.2% | estimated ± 2.1 pp, high confidence |
| 148 | GPT-5.4 | 0.2% | measured |
| 149 | Claude Opus 4.7 (Adaptive) | 0.2% | measured |
| 150 | GPT-5.6 Luna | 0.2% | measured |
| 151 | Qwen3.8 Max | 0.1% | estimated ± 2.1 pp, high confidence |
| 152 | GPT-6 Luna | 0.1% | measured |
| 153 | GPT-5.2 | 0.1% | estimated ± 2.1 pp, high confidence |
| 154 | Qwen3.7 Max | 0.1% | estimated ± 2.1 pp, high confidence |
| 155 | Pareto 26.10 Preview | 0.1% | estimated ± 3.8 pp, high confidence |
| 156 | Grok 4.20 | 0.1% | measured |
| 157 | Hy4 preview | 0.1% | estimated ± 2.1 pp, high confidence |
| 158 | Gemini 3.5 Flash | 0.1% | estimated ± 2.1 pp, medium confidence |
| 159 | Qwen3.8-Flash-Next | 0.0% | estimated ± 2.1 pp, medium confidence |
| 160 | Claude Opus 4.6 | 0.0% | estimated ± 2.1 pp, medium confidence |
| 161 | GLM-5.2 | 0.0% | estimated ± 2.1 pp, medium confidence |
| 162 | Qwen3.8-Omni-Flash | 0.0% | estimated ± 2.1 pp, medium confidence |
| 163 | DeepSeek V4.1 Flash | 0.0% | estimated ± 2.1 pp, medium confidence |
| 164 | Kimi K2.6 | 0.0% | estimated ± 2.1 pp, medium confidence |
| 165 | Beam | 0.0% | estimated ± 3.8 pp, high confidence |
| 166 | Qwen3.6 Plus | 0.0% | estimated ± 2.1 pp, medium confidence |
| 167 | Qwen3.7 Plus | 0.0% | estimated ± 2.1 pp, medium confidence |
| 168 | DeepSeek V4 Pro 0813 | 0.0% | estimated ± 2.1 pp, medium confidence |
| 169 | Grok 4.3 | 0.0% | estimated ± 2.1 pp, medium confidence |
| 170 | Claude Sonnet 4.6 | 0.0% | estimated ± 2.1 pp, medium confidence |
| 171 | Interfaze Beta | 0.0% | estimated ± 2.1 pp, medium confidence |
| 172 | Inkling-Small | 0.0% | estimated ± 2.1 pp, medium confidence |
| 173 | Ornith-1.5-35B-A3B | 0.0% | estimated ± 2.1 pp, medium confidence |
| 174 | Qwen3.8-27B | 0.0% | estimated ± 2.1 pp, medium confidence |
| 175 | Claude 3.5 Sonnet | 0.0% | estimated ± 2.1 pp, medium confidence |
| 176 | Claude Haiku 4.5 | 0.0% | estimated ± 9.2 pp, low confidence |
| 177 | Claude Opus 4.5 | 0.0% | estimated ± 2.1 pp, medium confidence |
| 178 | Claude Sonnet 4.5 | 0.0% | estimated ± 2.1 pp, medium confidence |
| 179 | Claude Sonnet 4.5 Thinking | 0.0% | estimated ± 13.8 pp, low confidence |
| 180 | DeepSeek V3 | 0.0% | estimated ± 2.1 pp, medium confidence |
| 181 | DeepSeek V4 Flash 0731 | 0.0% | estimated ± 2.1 pp, medium confidence |
| 182 | Gemini 2.5 Pro | 0.0% | estimated ± 2.1 pp, medium confidence |
| 183 | Gemini 3.1 Flash-Lite | 0.0% | estimated ± 9.2 pp, low confidence |
| 184 | Gemini 3 Pro Deep Think | 0.0% | estimated ± 13.8 pp, low confidence |
| 185 | Gemma 4 12B | 0.0% | estimated ± 2.1 pp, medium confidence |
| 186 | Gemma 4 31B | 0.0% | estimated ± 2.1 pp, medium confidence |
| 187 | Gemma 4 E2B | 0.0% | estimated ± 2.1 pp, medium confidence |
| 188 | Gemma 4 E4B | 0.0% | estimated ± 2.1 pp, medium confidence |
| 189 | GLM-4.5 | 0.0% | estimated ± 9.2 pp, low confidence |
| 190 | GLM-4.7 | 0.0% | estimated ± 2.1 pp, medium confidence |
| 191 | GLM-5 | 0.0% | estimated ± 2.1 pp, medium confidence |
| 192 | GLM-5.3-Flash | 0.0% | estimated ± 9.2 pp, low confidence |
| 193 | GPT-4.1 | 0.0% | estimated ± 2.1 pp, medium confidence |
| 194 | GPT-4.1 mini | 0.0% | estimated ± 2.1 pp, medium confidence |
| 195 | GPT-4.1 nano | 0.0% | estimated ± 2.1 pp, medium confidence |
| 196 | GPT-5.4 mini | 0.0% | estimated ± 2.1 pp, medium confidence |
| 197 | GPT-5.4 nano | 0.0% | estimated ± 2.1 pp, medium confidence |
| 198 | Granite 4.2 30B | 0.0% | estimated ± 2.1 pp, medium confidence |
| 199 | Granite 4.2 3B | 0.0% | estimated ± 2.1 pp, medium confidence |
| 200 | Granite 4.2 8B | 0.0% | estimated ± 2.1 pp, medium confidence |
| 201 | Hy3 Preview | 0.0% | estimated ± 2.1 pp, medium confidence |
| 202 | Inkling | 0.0% | estimated ± 2.1 pp, medium confidence |
| 203 | Kimi K2.5 | 0.0% | estimated ± 2.1 pp, medium confidence |
| 204 | Kimi K2.5 (Reasoning) | 0.0% | estimated ± 2.1 pp, medium confidence |
| 205 | Laguna M.1 | 0.0% | estimated ± 9.2 pp, low confidence |
| 206 | Laguna XS.2 | 0.0% | estimated ± 9.2 pp, low confidence |
| 207 | LFM2.5-230M | 0.0% | estimated ± 2.1 pp, medium confidence |
| 208 | LFM2.5-VL-450M | 0.0% | estimated ± 2.1 pp, medium confidence |
| 209 | Ling 2.6 Flash | 0.0% | estimated ± 2.1 pp, medium confidence |
| 210 | Ling 3.0 Flash | 0.0% | estimated ± 2.1 pp, medium confidence |
| 211 | Ling 3.0 Flash FP8 | 0.0% | estimated ± 2.1 pp, medium confidence |
| 212 | LLaDA2.2-mini | 0.0% | estimated ± 3.8 pp, medium confidence |
| 213 | LongCat-Flash-Lite-Sparse | 0.0% | estimated ± 3.8 pp, medium confidence |
| 214 | MAI-Thinking-1 | 0.0% | estimated ± 2.1 pp, medium confidence |
| 215 | Mellum2-12B-A2.5B-Instruct | 0.0% | estimated ± 2.1 pp, medium confidence |
| 216 | Mellum2-12B-A2.5B-Thinking | 0.0% | estimated ± 2.1 pp, medium confidence |
| 217 | MiMo-V2.5 | 0.0% | estimated ± 9.2 pp, low confidence |
| 218 | MiMo-V2-Flash | 0.0% | estimated ± 2.1 pp, medium confidence |
| 219 | MiniCPM5-1B | 0.0% | estimated ± 3.8 pp, medium confidence |
| 220 | Nemotron 3.5 Lightning 30B A3B NVFP4 | 0.0% | estimated ± 2.1 pp, medium confidence |
| 221 | Nemotron 3 Nano Omni 30B A3B | 0.0% | estimated ± 2.1 pp, medium confidence |
| 222 | Nemotron 3 Ultra | 0.0% | estimated ± 2.1 pp, medium confidence |
| 223 | o1 | 0.0% | estimated ± 2.1 pp, medium confidence |
| 224 | o1-pro | 0.0% | estimated ± 2.1 pp, medium confidence |
| 225 | o3-mini | 0.0% | estimated ± 2.1 pp, medium confidence |
| 226 | Ornith-1.5-9B | 0.0% | estimated ± 2.1 pp, medium confidence |
| 227 | Qwen3 235B 2507 | 0.0% | estimated ± 2.1 pp, medium confidence |
| 228 | Qwen3.5-122B-A10B | 0.0% | estimated ± 2.1 pp, medium confidence |
| 229 | Qwen3.5-27B | 0.0% | estimated ± 2.1 pp, medium confidence |
| 230 | Qwen3.5-35B-A3B | 0.0% | estimated ± 2.1 pp, medium confidence |
| 231 | Qwen3.5 397B | 0.0% | estimated ± 2.1 pp, medium confidence |
| 232 | Qwen3.5 Flash | 0.0% | estimated ± 9.2 pp, low confidence |
| 233 | Qwen3.6-27B | 0.0% | estimated ± 2.1 pp, medium confidence |
| 234 | Qwen3.6-35B-A3B | 0.0% | estimated ± 2.1 pp, medium confidence |
| 235 | Solar Open 2 | 0.0% | estimated ± 3.8 pp, medium confidence |
| 236 | Soofi S 30B-A3B | 0.0% | estimated ± 2.1 pp, medium confidence |
| 237 | Ternary Bonsai 2 27B | 0.0% | estimated ± 2.1 pp, medium confidence |
| 238 | ZAYA1-74B-Preview | 0.0% | estimated ± 2.1 pp, medium confidence |
| 239 | ZAYA1-8B | 0.0% | estimated ± 2.1 pp, medium confidence |
All leaderboards
Agentic · terminal
- AA Terminal-Bench 2.1
- AA Terminal-Bench 4.0
- Terminal-Bench 2.0
- Terminal-Bench 3.0
- Terminal-Bench 4.0
- Terminal-Bench-Science 0.1
- Terminal-Bench 2.1
- Terminal-Bench 2.1 (Vals AI)
Agentic · tools
- AA Agentic Index
- AA-AnalystAgent
- AA AutomationBench
- AA EnterpriseOps-Gym
- GDP.pdf
- AA Harvey LAB
- AA ITBench
- AA Tau3 Banking
- Agents' Last Exam
- APEX-Agents
- APEX-Agents-AA
- ApprenticeBench
- AutomationBench
- BFCL v4
- BrowseComp
- Claw-Eval
- CWE-bench v1
- CyberGym
- DeepPlanning
- DeepSearchQA
- DRACO
- ExploitGym
- GDPval-AA
- Gert Labs
- HLE w/ tools
- JobBench
- MCP Atlas
- MCP-Tasks
- OSWorld-Verified
- OSWorld 2.0
- QwenClawBench
- ResearchClawBench
- SkillsBench
- τ²-bench results
- τ³-bench results
- Toolathlon
- Toolathlon-Verified
- VITA-Bench
- WideResearch
Coding
- AA Coding Index
- AA-SciCode
- CursorBench 3.1
- CursorBench 3.2
- CursorBench 4.0
- DeepSWE
- FrontierCode 1.1 Main
- FrontierCode 1.1 Extended
- FrontierSWE v2
- LiveCodeBench
- LiveCodeBench Pro
- LiveCodeBench v6
- NL2Repo
- OpenHarmony Bench
- PostTrainBench v1.1
- ProgramBench
- React Native Evals
- SciCode
- SWE-bench Pro
- SWE-bench Verified
- SWE-Rebench
- Vals LiveCodeBench
- Vals SWE-bench
- Vibe Code Bench
- VulcanBench v3
Math
- AIME 2025
- AIME26
- FrontierMath (legacy)
- FrontierMath v2 (Tier 4)
- FrontierMath v2 (Tiers 1-3)
- HMMT Feb 2025
- HMMT Feb 2026
- HMMT Nov 2025
- IMOAnswerBench
- MATH-500
- MMAnswerBench
Knowledge & reasoning
- AA-GPQA Diamond
- AA-HLE
- ARC-AGI-2
- ARC-AGI-1
- ARC-AGI-3
- Artificial Analysis Intelligence Index
- BioMysteryBench (human-difficult)
- BioMysteryBench (human-solvable)
- C-Eval
- CritPt
- GPQA
- GPQA Diamond
- HealthBench (raw)
- HealthBench Hard
- HealthBench (length-adjusted)
- HealthBench Professional
- HealthBench Professional (raw)
- HLE
- HLE w/o tools
- HLE-Verified
- JevBench 1.4
- LABBench2
- MedXpertQA (Text)
- MMLU
- MMLU-Pro
- MMLU-Pro (Arcee)
- MMLU-Redux
- MMMLU
- AA-Omniscience Accuracy
- SuperGPQA
- Vals GPQA Diamond
- Vals MMLU-Pro
Instruction following
Multilingual
Vision & documents
- AA-MMMU-Pro
- BabyVision
- CharXiv
- CharXiv w/o tools
- ERQA
- MathVision
- MathVision w/ Python
- MedXpertQA (MM)
- MMMU
- MMMU-Pro
- MMMU-Pro w/ Python
- MMVU
- OfficeQA Pro
- OmniDocBench 1.5
- RealWorldQA
- RefCOCO (avg)
- ScreenSpot Pro
- SimpleVQA
- Video-MME (with subtitle)
- VideoMMMU
- V*
- ZeroBench