Knowledge & reasoning
HealthBench Professional leaderboard
As of 2026-11-21, the highest measured score on HealthBench Professional is 69.2% by Claude Sonnet 5.5. 204 more models have estimated scores, calibrated from the benchmarks they were measured on.
Measured scores: benchlm.ai.
| # | Model | Score | Source |
|---|---|---|---|
| 1 | Claude Sonnet 5.5 | 69.2% | measured |
| 2 | GPT-6.2 Astra | 66.3% | measured |
| 3 | GPT-6.2 Sol | 65.9% | measured |
| 4 | Gemini 4 Neon | 65.9% | estimated ± 4.2 pp, medium confidence |
| 5 | Claude Fable 5.1 | 65.8% | estimated ± 4.2 pp, medium confidence |
| 6 | Claude Opus 5.5 | 65.6% | measured |
| 7 | Ling 3.1 Flash | 65.4% | measured |
| 8 | Gemini 4 Argon | 64.8% | estimated ± 2.1 pp, medium confidence |
| 9 | GPT-6 Astra | 64.7% | measured |
| 10 | Gemini 4 Carbon | 64.6% | estimated ± 2.1 pp, medium confidence |
| 11 | Claude Mythos 5 | 64.5% | estimated ± 4.2 pp, medium confidence |
| 12 | GPT-6.1 Sol | 64.2% | measured |
| 13 | Claude Fable 5.2 | 63.7% | estimated ± 2.1 pp, medium confidence |
| 14 | GPT-6.1 Terra | 62.7% | measured |
| 15 | Muse Wonder Full | 62.3% | measured |
| 16 | Gemini 4 Hydrogen | 62.3% | estimated ± 4.2 pp, medium confidence |
| 17 | GPT-6 Luna | 60.8% | measured |
| 18 | GPT-6 Sol | 60.8% | measured |
| 19 | Claude Fable 5 | 60.5% | estimated ± 4.2 pp, medium confidence |
| 20 | GPT-5.6 Sol | 60.5% | measured |
| 21 | Muse Wonder | 60.5% | measured |
| 22 | Qwen4 Max | 60.0% | measured |
| 23 | Muse Spark 1.3 | 60.0% | estimated ± 4.2 pp, medium confidence |
| 24 | Claude Opus 5 | 59.8% | measured |
| 25 | Gemini 3.8 Flash | 59.4% | estimated ± 2.1 pp, medium confidence |
| 26 | Claude Haiku 5.5 | 59.4% | estimated ± 2.1 pp, medium confidence |
| 27 | Inkling 2 Small | 59.4% | estimated ± 2.1 pp, medium confidence |
| 28 | MiMo-V2.6-Pro | 59.4% | estimated ± 4.2 pp, medium confidence |
| 29 | Muse Spark 1.1 | 59.3% | measured |
| 30 | dots3-note Preview | 59.0% | estimated ± 4.3 pp, medium confidence |
| 31 | Kimi K3.1 Pro | 58.8% | estimated ± 4.2 pp, medium confidence |
| 32 | GPT-5.3 Codex | 58.2% | estimated ± 4.5 pp, medium confidence |
| 33 | Hy3 Preview | 58.1% | estimated ± 4.2 pp, medium confidence |
| 34 | Kimi K2.7 Code | 58.1% | estimated ± 4.5 pp, low confidence |
| 35 | Hy3 | 58.1% | estimated ± 4.5 pp, low confidence |
| 36 | Qwen3.5-122B-A10B | 58.1% | estimated ± 4.5 pp, low confidence |
| 37 | Qwen3.6-35B-A3B | 58.1% | estimated ± 4.5 pp, low confidence |
| 38 | Ling 3.0 Flash VL | 58.1% | estimated ± 4.5 pp, low confidence |
| 39 | o3 | 58.1% | estimated ± 4.5 pp, low confidence |
| 40 | Gemma 4 26B A4B | 58.1% | estimated ± 4.5 pp, low confidence |
| 41 | Quasar 438B | 58.1% | estimated ± 4.5 pp, low confidence |
| 42 | Trinity-Large-Preview | 58.1% | estimated ± 4.5 pp, low confidence |
| 43 | Trinity-Large-Thinking | 58.1% | estimated ± 4.5 pp, low confidence |
| 44 | Gemma 4 12B | 58.1% | estimated ± 4.5 pp, low confidence |
| 45 | Step 5 Preview | 57.9% | estimated ± 4.2 pp, medium confidence |
| 46 | Sakana Fugu-Ultra | 57.9% | estimated ± 4.2 pp, medium confidence |
| 47 | Muse Spark 1.2 | 57.8% | estimated ± 4.2 pp, medium confidence |
| 48 | Gemini 3.7 Flash | 57.8% | estimated ± 4.2 pp, medium confidence |
| 49 | GPT-5.6 Terra | 57.7% | measured |
| 50 | Muse Ember | 57.4% | estimated ± 4.2 pp, medium confidence |
| 51 | Gemini 3.6 Flash | 57.2% | estimated ± 4.2 pp, medium confidence |
| 52 | Pareto 26.9 | 57.1% | estimated ± 4.2 pp, medium confidence |
| 53 | Qwen3.8 Max Preview | 57.1% | estimated ± 4.2 pp, medium confidence |
| 54 | Grok 4.6 | 57.0% | estimated ± 3.6 pp, high confidence |
| 55 | Claude Opus 4.6 (Adaptive) | 57.0% | estimated ± 4.2 pp, medium confidence |
| 56 | GLM-5.3 | 57.0% | estimated ± 4.2 pp, medium confidence |
| 57 | Claude Opus 4.8 | 57.0% | estimated ± 3.6 pp, high confidence |
| 58 | Claude Opus 4.5 Thinking | 56.9% | estimated ± 4.2 pp, medium confidence |
| 59 | MiMo-V2-Pro | 56.9% | estimated ± 4.2 pp, medium confidence |
| 60 | GPT-5.2-Codex | 56.9% | estimated ± 4.2 pp, medium confidence |
| 61 | Qwen 3.6 Max (preview) | 56.9% | estimated ± 4.2 pp, medium confidence |
| 62 | Gemini 3 Pro | 56.8% | estimated ± 4.2 pp, medium confidence |
| 63 | GLM-5 | 56.8% | estimated ± 4.2 pp, medium confidence |
| 64 | MiMo-V2.6-Flash | 56.8% | estimated ± 4.2 pp, low confidence |
| 65 | Claude Opus 4.6 | 56.8% | estimated ± 3.6 pp, high confidence |
| 66 | GLM-5-Turbo | 56.8% | estimated ± 4.2 pp, low confidence |
| 67 | GPT-5.5 | 56.8% | estimated ± 3.6 pp, high confidence |
| 68 | Gemini 3.1 Pro | 56.8% | estimated ± 3.6 pp, high confidence |
| 69 | Mistral Large 4 | 56.8% | estimated ± 4.2 pp, low confidence |
| 70 | GLM-5.1 | 56.8% | estimated ± 4.2 pp, low confidence |
| 71 | Grok 4.5 | 56.8% | estimated ± 3.6 pp, high confidence |
| 72 | Muse Spark Open | 56.8% | estimated ± 4.2 pp, low confidence |
| 73 | GPT-5.1 | 56.8% | estimated ± 4.2 pp, low confidence |
| 74 | Claude Opus 4.7 | 56.8% | estimated ± 3.6 pp, high confidence |
| 75 | MiMo-V2-Omni | 56.8% | estimated ± 4.2 pp, low confidence |
| 76 | GPT-5.1-Codex | 56.8% | estimated ± 4.2 pp, low confidence |
| 77 | GPT-5.1-Codex-Max | 56.8% | estimated ± 4.2 pp, low confidence |
| 78 | Claude Opus 4.5 | 56.8% | estimated ± 4.2 pp, low confidence |
| 79 | Grok 4.20 | 56.8% | estimated ± 3.6 pp, medium confidence |
| 80 | GLM-5V-Turbo | 56.7% | estimated ± 4.2 pp, low confidence |
| 81 | Kimi K2.5 | 56.7% | estimated ± 4.2 pp, low confidence |
| 82 | GPT-5 | 56.7% | estimated ± 4.2 pp, low confidence |
| 83 | Qwen3.5-27B | 56.7% | estimated ± 4.2 pp, low confidence |
| 84 | MiniMax M2.7 | 56.7% | estimated ± 4.2 pp, low confidence |
| 85 | Command A+ | 56.7% | estimated ± 4.2 pp, low confidence |
| 86 | Grok 4 | 56.7% | estimated ± 4.2 pp, low confidence |
| 87 | GLM-4.7 | 56.7% | estimated ± 4.2 pp, low confidence |
| 88 | o3-pro | 56.7% | estimated ± 4.2 pp, low confidence |
| 89 | Claude 4 Opus Thinking | 56.7% | estimated ± 4.2 pp, low confidence |
| 90 | Grok 4.1 Fast (Reasoning) | 56.7% | estimated ± 4.2 pp, low confidence |
| 91 | Ling 3.0 Flash | 56.7% | estimated ± 4.2 pp, low confidence |
| 92 | Step 3.7 Flash | 56.7% | estimated ± 4.2 pp, low confidence |
| 93 | Qwen3.5-35B-A3B | 56.7% | estimated ± 4.2 pp, low confidence |
| 94 | MiniMax M3 | 56.7% | estimated ± 4.2 pp, low confidence |
| 95 | Claude 4.1 Opus | 56.7% | estimated ± 4.2 pp, low confidence |
| 96 | Grok 4 Fast (Reasoning) | 56.7% | estimated ± 4.2 pp, low confidence |
| 97 | Gemini 3 Flash | 56.7% | estimated ± 4.2 pp, low confidence |
| 98 | Claude 3.7 Sonnet Thinking | 56.7% | estimated ± 4.2 pp, low confidence |
| 99 | o4-mini | 56.7% | estimated ± 4.2 pp, low confidence |
| 100 | Claude 4 Opus | 56.7% | estimated ± 4.2 pp, low confidence |
| 101 | Claude 4 Sonnet | 56.7% | estimated ± 4.2 pp, low confidence |
| 102 | Gemini 2.5 Pro | 56.7% | estimated ± 4.2 pp, low confidence |
| 103 | MiMo-V2-Flash | 56.7% | estimated ± 4.2 pp, low confidence |
| 104 | DeepSeek V3.2 | 56.7% | estimated ± 4.2 pp, low confidence |
| 105 | Qwen3 Max | 56.7% | estimated ± 4.2 pp, low confidence |
| 106 | Claude 3.7 Sonnet | 56.7% | estimated ± 4.2 pp, low confidence |
| 107 | o1 | 56.7% | estimated ± 4.2 pp, low confidence |
| 108 | GLM-4.6 | 56.7% | estimated ± 4.2 pp, low confidence |
| 109 | Grok 4.7 | 56.7% | measured |
| 110 | K-Exaone | 56.7% | estimated ± 4.2 pp, low confidence |
| 111 | Grok Code Fast 1 | 56.7% | estimated ± 4.2 pp, low confidence |
| 112 | Ling 2.6 Flash | 56.7% | estimated ± 4.2 pp, low confidence |
| 113 | DeepSeek V3.1 | 56.7% | estimated ± 4.2 pp, low confidence |
| 114 | Apodex 1.1 | 56.7% | estimated ± 4.2 pp, low confidence |
| 115 | Apodex 1.1 Mini | 56.7% | estimated ± 4.2 pp, low confidence |
| 116 | DeepSeek-R1 | 56.7% | estimated ± 4.2 pp, low confidence |
| 117 | Kimi K2 | 56.7% | estimated ± 4.2 pp, low confidence |
| 118 | GPT-4.1 | 56.7% | estimated ± 4.2 pp, low confidence |
| 119 | o3-mini | 56.7% | estimated ± 4.2 pp, low confidence |
| 120 | o1-pro | 56.7% | estimated ± 4.2 pp, low confidence |
| 121 | Gemini 3.5 Flash-Lite | 56.7% | estimated ± 4.2 pp, low confidence |
| 122 | o1-preview | 56.7% | estimated ± 4.2 pp, low confidence |
| 123 | Grok 4.1 Fast | 56.7% | estimated ± 4.2 pp, low confidence |
| 124 | GLM-4.5-Air | 56.7% | estimated ± 4.2 pp, low confidence |
| 125 | GPT-4.1 mini | 56.7% | estimated ± 4.2 pp, low confidence |
| 126 | Gemini 2.5 Flash | 56.7% | estimated ± 4.2 pp, low confidence |
| 127 | o1-mini | 56.7% | estimated ± 4.2 pp, low confidence |
| 128 | DeepSeek V3 0324 | 56.7% | estimated ± 4.2 pp, low confidence |
| 129 | GPT-4.5 | 56.7% | estimated ± 4.2 pp, low confidence |
| 130 | QwQ-32B | 56.7% | estimated ± 4.2 pp, low confidence |
| 131 | Mistral Large 3 | 56.7% | estimated ± 4.2 pp, low confidence |
| 132 | Magistral Medium | 56.7% | estimated ± 4.2 pp, low confidence |
| 133 | Mistral Medium 3 | 56.7% | estimated ± 4.2 pp, low confidence |
| 134 | A.X K2 | 56.7% | estimated ± 4.2 pp, low confidence |
| 135 | Claude 3.5 Haiku | 56.7% | estimated ± 4.2 pp, low confidence |
| 136 | Gemini 2.0 Flash | 56.7% | estimated ± 4.2 pp, low confidence |
| 137 | Claude 3 Opus | 56.7% | estimated ± 4.2 pp, low confidence |
| 138 | DeepSeek V3 | 56.7% | estimated ± 4.2 pp, low confidence |
| 139 | GPT-4o | 56.7% | estimated ± 4.2 pp, low confidence |
| 140 | DeepSeek R1 Distill Qwen 32B | 56.7% | estimated ± 4.2 pp, low confidence |
| 141 | Magistral Small | 56.7% | estimated ± 4.2 pp, low confidence |
| 142 | Gemini 1.5 Pro | 56.7% | estimated ± 4.2 pp, low confidence |
| 143 | GPT-4.1 nano | 56.7% | estimated ± 4.2 pp, low confidence |
| 144 | Solar Pro 3 | 56.7% | estimated ± 4.2 pp, low confidence |
| 145 | Devstral Small | 56.7% | estimated ± 4.2 pp, low confidence |
| 146 | K-EXAONE 2.0 | 56.7% | estimated ± 4.2 pp, low confidence |
| 147 | Mistral Large 2 | 56.7% | estimated ± 4.2 pp, low confidence |
| 148 | Mistral Small 3.1 | 56.7% | estimated ± 4.2 pp, low confidence |
| 149 | Gemini 1.5 Flash | 56.7% | estimated ± 4.2 pp, low confidence |
| 150 | GPT-4 Turbo | 56.7% | estimated ± 4.2 pp, low confidence |
| 151 | Solar Pro 2 | 56.7% | estimated ± 4.2 pp, low confidence |
| 152 | Nova Pro | 56.7% | estimated ± 4.2 pp, low confidence |
| 153 | Qwen2.5 Coder 32B Instruct | 56.7% | estimated ± 4.2 pp, low confidence |
| 154 | Gemini 2.5 Flash-Lite | 56.7% | estimated ± 4.2 pp, low confidence |
| 155 | GPT-4o mini | 56.7% | estimated ± 4.2 pp, low confidence |
| 156 | DeepSeek-V2.5 | 56.7% | estimated ± 4.2 pp, low confidence |
| 157 | Muse Glimmer 30B | 56.7% | estimated ± 4.2 pp, low confidence |
| 158 | Claude 3 Haiku | 56.7% | estimated ± 4.2 pp, low confidence |
| 159 | Granite 4.2 30B | 56.7% | estimated ± 4.2 pp, low confidence |
| 160 | Nemotron 3.5 Lightning 30B A3B NVFP4 | 56.7% | estimated ± 4.2 pp, low confidence |
| 161 | Gemini 1.0 Pro | 56.7% | estimated ± 4.2 pp, low confidence |
| 162 | Granite 4.2 8B | 56.7% | estimated ± 4.2 pp, low confidence |
| 163 | Gemma 3 27B | 56.7% | estimated ± 4.2 pp, low confidence |
| 164 | Gemma 4 E4B | 56.7% | estimated ± 4.2 pp, low confidence |
| 165 | Granite 4.2 3B | 56.7% | estimated ± 4.2 pp, low confidence |
| 166 | Mistral Medium 3.5 128B | 56.7% | estimated ± 4.2 pp, low confidence |
| 167 | Mistral Small 4 | 56.7% | estimated ± 4.2 pp, low confidence |
| 168 | GPT-OSS 120B | 56.7% | estimated ± 4.2 pp, low confidence |
| 169 | North Mini Code | 56.7% | estimated ± 4.2 pp, low confidence |
| 170 | GPT-5.2 | 55.8% | estimated ± 4.2 pp, medium confidence |
| 171 | Pareto 26.10 Preview | 55.8% | estimated ± 4.2 pp, medium confidence |
| 172 | Qwen3.7 Max | 55.8% | estimated ± 4.2 pp, medium confidence |
| 173 | Sakana Fugu | 55.7% | estimated ± 4.2 pp, medium confidence |
| 174 | GPT-5.6 Luna | 55.7% | measured |
| 175 | Gemini 3.5 Flash | 55.4% | estimated ± 4.2 pp, low confidence |
| 176 | ERNIE 5.5 Flash Preview | 55.1% | estimated ± 4.2 pp, medium confidence |
| 177 | Ling 3.0 Flash Sante | 54.3% | measured |
| 178 | Ornith-1.5-397B | 53.6% | estimated ± 4.2 pp, medium confidence |
| 179 | Qwen3.8-Omni-Flash | 53.1% | estimated ± 4.2 pp, low confidence |
| 180 | DeepSeek V4.1 Flash 1017 | 52.9% | estimated ± 4.2 pp, low confidence |
| 181 | Qwen3.8 Max | 52.7% | estimated ± 4.2 pp, medium confidence |
| 182 | Kimi K3 | 52.7% | estimated ± 4.2 pp, medium confidence |
| 183 | Hy4 preview | 52.6% | estimated ± 4.2 pp, medium confidence |
| 184 | MiniMax M3.1 Flash | 52.6% | estimated ± 4.2 pp, medium confidence |
| 185 | Claude Sonnet 5 | 52.4% | estimated ± 4.2 pp, medium confidence |
| 186 | GPT-5.5 Pro | 52.3% | estimated ± 4.2 pp, medium confidence |
| 187 | Kimi K2.6 | 52.1% | estimated ± 4.2 pp, low confidence |
| 188 | Muse Spark | 52.1% | estimated ± 4.2 pp, medium confidence |
| 189 | GPT-5.4 Pro | 52.0% | estimated ± 4.2 pp, medium confidence |
| 190 | Particle-1 | 51.9% | estimated ± 4.2 pp, low confidence |
| 191 | Qwen3.6 Plus | 51.9% | estimated ± 4.2 pp, low confidence |
| 192 | Qwen3.7 Plus | 51.7% | estimated ± 4.2 pp, low confidence |
| 193 | DeepSeek V4 Pro | 51.3% | estimated ± 4.2 pp, low confidence |
| 194 | Grok 4.3 | 51.3% | estimated ± 4.2 pp, low confidence |
| 195 | Claude Sonnet 4.6 | 50.9% | estimated ± 4.2 pp, low confidence |
| 196 | Interfaze Beta | 50.9% | estimated ± 4.2 pp, low confidence |
| 197 | GLM-5.2 | 50.1% | estimated ± 4.2 pp, medium confidence |
| 198 | Mercury 3 | 49.6% | estimated ± 4.2 pp, low confidence |
| 199 | Solar Pro 4 | 49.2% | estimated ± 4.2 pp, low confidence |
| 200 | GPT-5.4 | 48.1% | measured |
| 201 | Qwen3.5 397B | 48.0% | estimated ± 4.2 pp, low confidence |
| 202 | OrcaCyber-Nano-1.0 | 47.8% | estimated ± 4.2 pp, low confidence |
| 203 | DeepSeek V4 Flash 0731 | 47.4% | estimated ± 4.2 pp, low confidence |
| 204 | Qwen3.6-27B | 46.8% | estimated ± 4.2 pp, low confidence |
| 205 | Beam 1.0 501B-A23B | 46.0% | estimated ± 4.2 pp, low confidence |
| 206 | Beam 1.0 501B-A23B FP8 | 46.0% | estimated ± 4.2 pp, low confidence |
| 207 | Qwen3.8-Flash-Next | 45.9% | estimated ± 4.2 pp, low confidence |
| 208 | Beam 1.0 501B-A23B NVFP4 | 44.8% | estimated ± 4.2 pp, low confidence |
| 209 | MiMo-V2.5-Pro | 44.1% | estimated ± 4.2 pp, low confidence |
| 210 | Nemotron 3.5 Super VL 120B A12B | 41.9% | estimated ± 4.2 pp, low confidence |
| 211 | Inkling-Small | 41.7% | estimated ± 4.2 pp, low confidence |
| 212 | Nemotron 3.5 Super VL 120B A12B NVFP4 | 41.5% | estimated ± 4.2 pp, low confidence |
| 213 | Qwen3.8-27B | 40.9% | estimated ± 4.2 pp, low confidence |
| 214 | Inkling 2 Mini | 40.8% | estimated ± 4.2 pp, low confidence |
| 215 | Inkling | 40.1% | estimated ± 4.2 pp, low confidence |
| 216 | Solar Open 2 | 38.8% | estimated ± 4.2 pp, low confidence |
| 217 | Nemotron 3.5 Bolt 74B-A9B BF16 | 38.7% | estimated ± 4.2 pp, low confidence |
| 218 | GPT-5.4 mini | 38.2% | estimated ± 4.2 pp, low confidence |
| 219 | Nemotron 3 Ultra | 36.6% | estimated ± 4.2 pp, low confidence |
| 220 | Ornith-1.5-35B-A3B | 35.4% | estimated ± 4.2 pp, low confidence |
| 221 | Nemotron 3.5 Bolt 74B-A9B | 35.0% | estimated ± 4.2 pp, low confidence |
| 222 | GPT-5.4 nano | 33.9% | estimated ± 4.2 pp, low confidence |
| 223 | Ornith-1.5-9B | 29.2% | estimated ± 4.2 pp, low confidence |
| 224 | Gemma 4 31B | 28.3% | estimated ± 4.2 pp, low confidence |
| 225 | Spark-X2.6-4B | 26.3% | estimated ± 4.2 pp, low confidence |
All leaderboards
Agentic · terminal
- AA Terminal-Bench 2.1
- AA Terminal-Bench 4.0
- Terminal-Bench 4.0
- Terminal-Bench 2.0
- Terminal-Bench 2.1
- Terminal-Bench 3.0
- Terminal-Bench Hard
- Terminal-Bench-Science 0.1
- Terminal-Bench 4.1
- Terminal-Bench 2.1 (Vals AI)
Agentic · tools
- AA Agentic Index
- AA-AnalystAgent
- AA AutomationBench
- CWE-Bench-AA
- AA Cyber Index
- DeepsecBench-AA
- AA EnterpriseOps-Gym
- GDP.pdf
- AA Harvey LAB v1.0
- Harvey LAB-AA v1.1
- AA ITBench
- AA Tau3 Banking
- Agents' Last Exam
- APEX-Agents
- APEX-Agents-AA
- ApprenticeBench
- AutomationBench
- BFCL v4
- BrowseComp
- Claw-Eval
- CWE-bench v1
- CWE-bench
- CyberGym
- DeepPlanning
- DeepSearchQA
- DRACO
- ExploitGym
- Finance Agent v2
- GDPval-AA
- Gert Labs
- Hermes Bench
- Hermes Index
- HLE w/ tools
- JobBench
- LifeVal
- MazeBench
- MCP Atlas
- MCP-Tasks
- MKA
- OSWorld 2.1
- OSWorld 2.1 Offline
- OSWorld-Verified
- OSWorld 2.0
- PinchBench
- QwenClawBench
- ResearchClawBench
- SkillsBench
- τ²-bench results
- τ³-bench results
- Toolathlon
- Toolathlon-Verified
- VITA-Bench
- WideResearch
Coding
- AA Coding Index
- AA LiveCodeBench
- AA-SciCode
- Bug Hunt Bench
- CursorBench 3.1
- CursorBench 3.2
- CursorBench 4.0
- cursorBench41
- DeepSWE
- FrontierCode 1.1 Main
- FrontierCode 2.0
- FrontierCode 1.1 Extended
- FrontierSWE v2
- LiveCodeBench
- LiveCodeBench Pro
- LiveCodeBench v6
- NL2Repo
- OpenHarmony Bench v1.2
- OpenHarmony Bench v1.0
- PostTrainBench
- PostTrainBench v1
- PostTrainBench v1.1
- ProgramBench
- React Native Evals
- ScarfBench
- SciCode
- SWE-Atlas Codebase QnA
- SWE-bench Pro
- SWE-bench Verified
- SWE-Rebench
- SWE-Marathon
- Vals LiveCodeBench
- Vals SWE-bench
- Vibe Code Bench
- VulcanBench Frontier v4
- VulcanBench v3
Math
- AIME 2024
- AIME 2025
- AIME26
- 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
- HealthBench Professional (length-adjusted, Artificial Analysis)
- HealthBench Professional (raw, Artificial Analysis)
- AA-HLE
- AA MMLU-Pro
- ARC-AGI-2
- ARC-AGI-3 Provider Adapter
- ARC-AGI-1
- ARC-AGI-3
- Artificial Analysis Intelligence Index
- BBH
- BioBasicsBench
- 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
- SimpleQA
- STEMBench
- SuperGPQA
- Vals GPQA Diamond
- Vals MMLU-Pro
Instruction following
Multilingual
Vision & documents
- AA-MMMU-Pro
- AI2D_TEST
- BabyVision
- BabyVision w/ Python
- BenchCAD Vision2Code (no tools)
- BenchCAD Vision2Code (tools)
- CC-OCR
- Chartography (no tools)
- Chartography (tools)
- CharXiv
- CharXiv w/o tools
- CountBench
- DynaMath
- ERQA
- LVBench
- MathVision
- MathVision w/ Python
- MedXpertQA (MM)
- MLVU (M-Avg)
- MMMU
- MMMU-Pro
- MMMU-Pro w/ Python
- MMVU
- OCRBench V2
- OfficeQA
- OfficeQA Pro
- OmniDocBench 1.5
- ParseBench-1
- ParseBench-2
- PerceptionBench
- RealWorldQA
- RefCOCO (avg)
- ScreenSpot Pro
- SimpleVQA
- SWE-bench Multimodal
- Video-MME (w/o subtitle)
- Video-MME (with subtitle)
- VideoMMMU
- V*
- We-Math
- ZeroBench
- ZeroBench w/ Python
Long context
General
- HLE-MultiChoice (Decision Index)
- Artificial Analysis Decision Index v1.0
- Artificial Analysis Decision Index
- When2Call (Decision Index)
- WildGuardTest (Decision Index)
- BabaIsBench
- BinaryAudit
- BlenderWorld (text-only)
- Chess Puzzles
- CompileBench
- Cookie-Bench
- DAYJOB
- DAYJOB Healthcare
- Decision Index Full Score
- EBR-bench
- ExploitBench
- ExploitBench Internal Port
- FACTS Multimodal
- FailureSensorIQ
- FrontierPhysics
- Furniture Assembly
- Decision Index Arts
- Decision Index Knowledge
- Decision Index Language
- Decision Index Retrieval
- Decision Index Tools
- MaxxBench
- MirrorCode
- Mystery Game Puzzles
- NYT Connections
- OTelBench
- SEC-Bench Pro
- SkillsBench official leaderboard (BenchFlow)
- SpireClimb
- StealthMark