MedQA Bias Evaluation
Models answer medical exam questions before and after Vals adds racial information. Vals compares accuracy across patient groups.
- Made by
- Vals AI
- Released
- January 2025
- Kind
- Medicine
- Level
- Expert
- Scored
- Checked automatically
- Answers
- Multiple choice
- Measures
- Propensity
- Risk
- Specialized advice
- Behavior
- Bias
- Subject
- Clinical medicine
- Language
- English
- Interaction
- Single turn
- Items from
- Exams and contests, Another benchmark
- Refresh
- One fixed set
- Run by
- A host
- Based on
- MedQA
- In views
- Biology and medicine, Content safety
- Strongest model tested
- GPT-5.4 #28 on the capability index
- Last model added
- March 2026
- Scores refreshed
- November 20th, 2026
o1 leads with 96.5% among models released by March 2026 o1 leads with 96.5% among models released by March 2026
Best score per entry from Vals AI’s leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.
The newest model with a score here came out in March 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
- 96.5 o1
- 96.4 GPT-5.1
- 96.4
Gemini 3.1 Pro
- 96.3 GPT-5
- 96.1 GPT-5.4
- 96.1 o3
- 96.1 GPT-5 mini
- 96.0
Gemini 3 Pro
- 96.0 o4-mini
- 95.9 Claude Opus 4.5
- 95.8
Gemini 3 Flash
- 95.4 Claude Opus 4.6
- 95.2
Qwen 3.5 Plus (hosted 397B-A17B)
- 94.8 o3-mini
- 94.7 Claude Sonnet 4.5
- 94.5 Grok 4.20
- 94.4 Kimi K2.5
- 94.3 GLM-5
- 94.1 GPT-5.2
- 93.9
DeepSeek-V3.2
- 93.7 GLM-4.7
- 93.6 Claude Opus 4.1
- 93.3 GPT-5 nano
- 93.1
Gemini 2.5 Pro
- 93.0 o1-preview
- 92.9 Claude Opus 4
- 92.7 Claude Sonnet 4
- 92.6 Kimi K2 Thinking
- 92.5
MiniMax-M2.5
- 92.5 Grok 4
- 92.3 Grok-2
- 92.2 GLM-4.6
- 92.1 Grok 4.1 Fast
- 92.1 Grok 4 Fast
- 92.1 Claude Sonnet 4.6
- 91.4
Gemini 2.5 Flash
- 91.4 gpt-oss-120b
- 91.2 GPT-4.1
- 91.2
MiniMax-M2.1
- 91.0
Gemini 2.5 Flash
- 90.8
DeepSeek-R1
- 90.6
Qwen3-235B-A22B
- 90.2 o1-mini
- 90.2 Claude 3.7 Sonnet
- 90.1 Grok-3 mini
- 90.0 GLM-4.5
- 89.5
Magistral Medium 1.2
- 88.9
Gemini 2.5 Flash-Lite
- 88.2
Llama 3.1-405B
- 88.2 GPT-4o
- 87.4
Qwen3-Max-Preview
- 87.4
Qwen3-Max
- 84.8
Llama 3.1-70B
- 84.6 GPT-4.1 mini
- 84.0 Kimi K2
- 83.8 Grok 3
- 83.2 Claude 3.5 Sonnet
- 82.9 gpt-oss-20b
- 82.4
Magistral Small 1.2
- 82.2
Mistral Large 3
- 82.0
DeepSeek-V3
- 82.0 GPT-4 Turbo
- 81.5
Gemini 2.0 Flash
- 80.5
Cohere Command A
- 79.6 Claude Haiku 4.5
- 78.2
Mistral Medium 3.1
- 77.4
Qwen2.5-72B
- 76.5
Gemini 1.5 Pro
- 76.2
Mistral Large 2
- 72.4 GPT-4o mini
- 69.1
Mistral Small 3.1
- 68.2 GPT-4.1 nano
- 68.1 Jamba 1.5 Large
- 62.6
Llama 3.1-8B
- 62.1
Mixtral 8x22B
- 58.5 GPT 3.5
- 57.0 mistral-small-2402
- 55.2 Jamba 1.5 Mini
- 53.2
Mixtral 8x7B
- 52.5 Jamba 1.6 Mini
- 50.9
Llama 4 Scout
- 50.7 Jamba 1.6 Large
- 43.3
Llama 4 Maverick
- 2.65
Command R+
- 1 o1 96.5%
- 2 GPT-5.1 96.4%
- 3
Gemini 3.1 Pro 96.4%
- 4 GPT-5 96.3%
- 5 GPT-5.4 xhigh 96.1%
- 6 o3 96.1%
- 7 GPT-5 mini 96.1%
- 8
Gemini 3 Pro 96%
- 9 o4-mini 96%
- 10 Claude Opus 4.5 thinking 95.9%
- 11
Gemini 3 Flash 95.8%
- 12 Claude Opus 4.6 thinking 95.4%
- 13
Qwen 3.5 Plus (hosted 397B-A17B) thinking 95.2%
- 14 o3-mini 94.8%
- 15 Claude Sonnet 4.5 thinking 94.7%
- 16 Grok 4.20 reasoning 94.5%
- 17 Kimi K2.5 thinking 94.4%
- 18 GLM-5 thinking 94.3%
- 19 GPT-5.2 94.1%
- 20
DeepSeek-V3.2 thinking 93.9%
- 21 GLM-4.7 93.7%
- 22 Claude Opus 4.1 thinking 93.6%
- 23 GPT-5 nano 93.3%
- 24
Gemini 2.5 Pro Jun 2025 93.1%
- 25 o1-preview 93%
- 26 Claude Opus 4 non-thinking 92.9%
- 27 Claude Sonnet 4 thinking 92.7%
- 28 Kimi K2 Thinking 92.6%
- 29
MiniMax-M2.5 92.5%
- 30 Grok 4 92.5%
- 31 Grok-2 Dec 2024 92.3%
- 32 GLM-4.6 92.2%
- 33 Grok 4.1 Fast reasoning 92.1%
- 34 Grok 4 Fast reasoning 92.1%
- 35 Claude Sonnet 4.6 92.1%
- 36
Gemini 2.5 Flash Sep 2025, non-thinking 91.4%
- 37 gpt-oss-120b 91.4%
- 38 GPT-4.1 91.2%
- 39
MiniMax-M2.1 91.2%
- 40
Gemini 2.5 Flash Apr 2025, thinking 91%
- 41
DeepSeek-R1 May 2025 90.8%
- 42
Qwen3-235B-A22B 90.6%
- 43 o1-mini 90.2%
- 44 Claude 3.7 Sonnet thinking 90.2%
- 45 Grok-3 mini reasoning 90.1%
- 46 GLM-4.5 90%
- 47
Magistral Medium 1.2 89.5%
- 48
Gemini 2.5 Flash-Lite Sep 2025, thinking 88.9%
- 49
Llama 3.1-405B 88.2%
- 50 GPT-4o Aug 2024 88.2%
- 51
Qwen3-Max-Preview 87.4%
- 52
Qwen3-Max 87.4%
- 53
Llama 3.1-70B 84.8%
- 54 GPT-4.1 mini 84.6%
- 55 Kimi K2 Jul 2025 84%
- 56 Grok 3 83.8%
- 57 Claude 3.5 Sonnet 83.2%
- 58 gpt-oss-20b 82.9%
- 59
Magistral Small 1.2 82.4%
- 60
Mistral Large 3 82.2%
- 61
DeepSeek-V3 Mar 2025 82%
- 62 GPT-4 Turbo Apr 2024 82%
- 63
Gemini 2.0 Flash Feb 2025 81.5%
- 64
Cohere Command A 80.5%
- 65 Claude Haiku 4.5 thinking 79.6%
- 66
Mistral Medium 3.1 78.2%
- 67
Qwen2.5-72B 77.4%
- 68
Gemini 1.5 Pro Sep 2024 76.5%
- 69
Mistral Large 2 Nov 2024 76.2%
- 70 GPT-4o mini 72.4%
- 71
Mistral Small 3.1 69.1%
- 72 GPT-4.1 nano 68.2%
- 73 Jamba 1.5 Large 68.1%
- 74
Llama 3.1-8B 62.6%
- 75
Mixtral 8x22B 62.1%
- 76 GPT 3.5 58.5%
- 77 mistral-small-2402 57%
- 78 Jamba 1.5 Mini 55.2%
- 79
Mixtral 8x7B 53.2%
- 80 Jamba 1.6 Mini 52.5%
- 81
Llama 4 Scout 50.9%
- 82 Jamba 1.6 Large 50.7%
- 83
Llama 4 Maverick 43.3%
- 84
Command R+ 2.65%
Source: Vals AI’s leaderboard
- 96.5 o1
- 96.4 GPT-5.1
- 96.4
Gemini 3.1 Pro
- 96.3 GPT-5
- 96.1 GPT-5.4
- 96.1 o3
- 96.1 GPT-5 mini
- 96.0
Gemini 3 Pro
- 96.0 o4-mini
- 95.9 Claude Opus 4.5
- 95.8
Gemini 3 Flash
- 95.4 Claude Opus 4.6
- 95.2
Qwen 3.5 Plus (hosted 397B-A17B)
- 94.8 o3-mini
- 94.7 Claude Sonnet 4.5
- 94.5 Grok 4.20
- 94.4 Kimi K2.5
- 94.3 GLM-5
- 94.1 GPT-5.2
- 93.9
DeepSeek-V3.2
- 93.7 GLM-4.7
- 93.6 Claude Opus 4.1
- 93.3 GPT-5 nano
- 93.1
Gemini 2.5 Pro
- 93.0 o1-preview
- 92.9 Claude Opus 4
- 92.7 Claude Sonnet 4
- 92.6 Kimi K2 Thinking
- 92.5
MiniMax-M2.5
- 92.5 Grok 4
- 92.3 Grok-2
- 92.2 GLM-4.6
- 92.1 Grok 4.1 Fast
- 92.1 Grok 4 Fast
- 92.1 Claude Sonnet 4.6
- 91.4
Gemini 2.5 Flash
- 91.4 gpt-oss-120b
- 91.2 GPT-4.1
- 91.2
MiniMax-M2.1
- 91.0
Gemini 2.5 Flash
- 90.8
DeepSeek-R1
- 90.6
Qwen3-235B-A22B
- 90.2 o1-mini
- 90.2 Claude 3.7 Sonnet
- 90.1 Grok-3 mini
- 90.0 GLM-4.5
- 89.5
Magistral Medium 1.2
- 88.9
Gemini 2.5 Flash-Lite
- 88.2
Llama 3.1-405B
- 88.2 GPT-4o
- 87.4
Qwen3-Max-Preview
- 87.4
Qwen3-Max
- 84.8
Llama 3.1-70B
- 84.6 GPT-4.1 mini
- 84.0 Kimi K2
- 83.8 Grok 3
- 83.2 Claude 3.5 Sonnet
- 82.9 gpt-oss-20b
- 82.4
Magistral Small 1.2
- 82.2
Mistral Large 3
- 82.0
DeepSeek-V3
- 82.0 GPT-4 Turbo
- 81.5
Gemini 2.0 Flash
- 80.5
Cohere Command A
- 79.6 Claude Haiku 4.5
- 78.2
Mistral Medium 3.1
- 77.4
Qwen2.5-72B
- 76.5
Gemini 1.5 Pro
- 76.2
Mistral Large 2
- 72.4 GPT-4o mini
- 69.1
Mistral Small 3.1
- 68.2 GPT-4.1 nano
- 68.1 Jamba 1.5 Large
- 62.6
Llama 3.1-8B
- 62.1
Mixtral 8x22B
- 57.0 mistral-small-2402
- 55.2 Jamba 1.5 Mini
- 53.2
Mixtral 8x7B
- 52.5 Jamba 1.6 Mini
- 50.9
Llama 4 Scout
- 50.7 Jamba 1.6 Large
- 43.3
Llama 4 Maverick
- 2.65
Command R+
- 1 o1 96.5%
- 2 GPT-5.1 96.4%
- 3
Gemini 3.1 Pro 96.4%
- 4 GPT-5 96.3%
- 5 GPT-5.4 xhigh 96.1%
- 6 o3 96.1%
- 7 GPT-5 mini 96.1%
- 8
Gemini 3 Pro 96%
- 9 o4-mini 96%
- 10 Claude Opus 4.5 thinking 95.9%
- 11
Gemini 3 Flash 95.8%
- 12 Claude Opus 4.6 thinking 95.4%
- 13
Qwen 3.5 Plus (hosted 397B-A17B) thinking 95.2%
- 14 o3-mini 94.8%
- 15 Claude Sonnet 4.5 thinking 94.7%
- 16 Grok 4.20 reasoning 94.5%
- 17 Kimi K2.5 thinking 94.4%
- 18 GLM-5 thinking 94.3%
- 19 GPT-5.2 94.1%
- 20
DeepSeek-V3.2 thinking 93.9%
- 21 GLM-4.7 93.7%
- 22 Claude Opus 4.1 thinking 93.6%
- 23 GPT-5 nano 93.3%
- 24
Gemini 2.5 Pro Jun 2025 93.1%
- 25 o1-preview 93%
- 26 Claude Opus 4 non-thinking 92.9%
- 27 Claude Sonnet 4 thinking 92.7%
- 28 Kimi K2 Thinking 92.6%
- 29
MiniMax-M2.5 92.5%
- 30 Grok 4 92.5%
- 31 Grok-2 Dec 2024 92.3%
- 32 GLM-4.6 92.2%
- 33 Grok 4.1 Fast reasoning 92.1%
- 34 Grok 4 Fast reasoning 92.1%
- 35 Claude Sonnet 4.6 92.1%
- 36
Gemini 2.5 Flash Sep 2025, non-thinking 91.4%
- 37 gpt-oss-120b 91.4%
- 38 GPT-4.1 91.2%
- 39
MiniMax-M2.1 91.2%
- 40
Gemini 2.5 Flash Apr 2025, thinking 91%
- 41
DeepSeek-R1 May 2025 90.8%
- 42
Qwen3-235B-A22B 90.6%
- 43 o1-mini 90.2%
- 44 Claude 3.7 Sonnet thinking 90.2%
- 45 Grok-3 mini reasoning 90.1%
- 46 GLM-4.5 90%
- 47
Magistral Medium 1.2 89.5%
- 48
Gemini 2.5 Flash-Lite Sep 2025, thinking 88.9%
- 49
Llama 3.1-405B 88.2%
- 50 GPT-4o Aug 2024 88.2%
- 51
Qwen3-Max-Preview 87.4%
- 52
Qwen3-Max 87.4%
- 53
Llama 3.1-70B 84.8%
- 54 GPT-4.1 mini 84.6%
- 55 Kimi K2 Jul 2025 84%
- 56 Grok 3 83.8%
- 57 Claude 3.5 Sonnet 83.2%
- 58 gpt-oss-20b 82.9%
- 59
Magistral Small 1.2 82.4%
- 60
Mistral Large 3 82.2%
- 61
DeepSeek-V3 Mar 2025 82%
- 62 GPT-4 Turbo Apr 2024 82%
- 63
Gemini 2.0 Flash Feb 2025 81.5%
- 64
Cohere Command A 80.5%
- 65 Claude Haiku 4.5 thinking 79.6%
- 66
Mistral Medium 3.1 78.2%
- 67
Qwen2.5-72B 77.4%
- 68
Gemini 1.5 Pro Sep 2024 76.5%
- 69
Mistral Large 2 Nov 2024 76.2%
- 70 GPT-4o mini 72.4%
- 71
Mistral Small 3.1 69.1%
- 72 GPT-4.1 nano 68.2%
- 73 Jamba 1.5 Large 68.1%
- 74
Llama 3.1-8B 62.6%
- 75
Mixtral 8x22B 62.1%
- 76 mistral-small-2402 57%
- 77 Jamba 1.5 Mini 55.2%
- 78
Mixtral 8x7B 53.2%
- 79 Jamba 1.6 Mini 52.5%
- 80
Llama 4 Scout 50.9%
- 81 Jamba 1.6 Large 50.7%
- 82
Llama 4 Maverick 43.3%
- 83
Command R+ 2.65%
Source: Vals AI’s leaderboard
| # | Entry | Score ± | Setting | Released | Run on | Source |
|---|---|---|---|---|---|---|
| 1 | o1 | 96.5% ±0.6 | Dec 2024 | Official leaderboard | ||
| 2 | GPT-5.1 | 96.4% ±0.3 | Nov 2025 | Official leaderboard | ||
| 3 | | 96.4% ±0.3 | Feb 2026 | Official leaderboard | ||
| 4 | GPT-5 | 96.3% ±0.3 | Aug 2025 | Official leaderboard | ||
| 5 | GPT-5.4 | 96.1% ±0.3 | xhigh | Mar 2026 | Official leaderboard | |
| 6 | o3 | 96.1% ±0.3 | Apr 2025 | Official leaderboard | ||
| 7 | GPT-5 mini | 96.1% ±0.3 | Aug 2025 | Official leaderboard | ||
| 8 | | 96% ±0.3 | Nov 2025 | Official leaderboard | ||
| 9 | o4-mini | 96% ±0.3 | Apr 2025 | Official leaderboard | ||
| 10 | Claude Opus 4.5 | 95.9% ±0.4 | thinking | Nov 2025 | Official leaderboard | |
| 11 | | 95.8% ±0.4 | Dec 2025 | Official leaderboard | ||
| 12 | Claude Opus 4.6 | 95.4% ±0.4 | thinking | Feb 2026 | Official leaderboard | |
| 13 | | 95.2% ±0.4 | thinking | Feb 2026 | Official leaderboard | |
| 14 | o3-mini | 94.8% ±0.8 | Jan 2025 | Official leaderboard | ||
| 15 | Claude Sonnet 4.5 | 94.7% ±0.4 | thinking | Sep 2025 | Official leaderboard | |
| 16 | Grok 4.20 | 94.5% ±0.4 | reasoning | Feb 2026 | Official leaderboard | |
| 17 | Kimi K2.5 | 94.4% ±0.4 | thinking | Jan 2026 | Official leaderboard | |
| 18 | GLM-5 | 94.3% ±0.4 | thinking | Feb 2026 | Official leaderboard | |
| 19 | GPT-5.2 | 94.1% ±0.4 | Dec 2025 | Official leaderboard | ||
| 20 | | 93.9% ±0.4 | thinking | Dec 2025 | Official leaderboard | |
| 21 | GLM-4.7 | 93.7% ±0.4 | Dec 2025 | Official leaderboard | ||
| 22 | Claude Opus 4.1 | 93.6% ±0.4 | thinking | Aug 2025 | Official leaderboard | |
| 23 | GPT-5 nano | 93.3% ±0.4 | Aug 2025 | Official leaderboard | ||
| 24 | | 93.1% ±0.5 | Jun 2025 | Official leaderboard | ||
| 25 | o1-preview | 93% ±0.9 | Sep 2024 | Official leaderboard | ||
| 26 | Claude Opus 4 | 92.9% ±0.5 | non-thinking | May 2025 | Official leaderboard | |
| 27 | Claude Sonnet 4 | 92.7% ±0.5 | thinking | May 2025 | Official leaderboard | |
| 28 | Kimi K2 Thinking | 92.6% ±0.5 | Nov 2025 | Official leaderboard | ||
| 29 | | 92.5% ±0.5 | Feb 2026 | Official leaderboard | ||
| 30 | Grok 4 | 92.5% ±0.5 | Jul 2025 | Official leaderboard | ||
| 31 | Grok-2 Dec 2024 | 92.3% ±0.9 | Dec 2024 | Official leaderboard | ||
| 32 | GLM-4.6 | 92.2% ±0.5 | Sep 2025 | Official leaderboard | ||
| 33 | Grok 4.1 Fast | 92.1% ±0.5 | reasoning | Nov 2025 | Official leaderboard | |
| 34 | Grok 4 Fast | 92.1% ±0.5 | reasoning | Sep 2025 | Official leaderboard | |
| 35 | Claude Sonnet 4.6 | 92.1% ±0.5 | Feb 2026 | Official leaderboard | ||
| 36 | | 91.4% ±0.5 | non-thinking | Sep 2025 | Official leaderboard | |
| 37 | gpt-oss-120b | 91.4% ±0.5 | Aug 2025 | Official leaderboard | ||
| 38 | GPT-4.1 | 91.2% ±0.5 | Apr 2025 | Official leaderboard | ||
| 39 | | 91.2% ±0.5 | Dec 2025 | Official leaderboard | ||
| 40 | | 91% ±0.5 | thinking | Apr 2025 | Official leaderboard | |
| 41 | | 90.8% ±1 | May 2025 | Official leaderboard | ||
| 42 | | 90.6% ±0.5 | Apr 2025 | Official leaderboard | ||
| 43 | o1-mini | 90.2% | Sep 2024 | Official leaderboard | ||
| 44 | Claude 3.7 Sonnet | 90.2% ±1 | thinking | Feb 2025 | Official leaderboard | |
| 45 | Grok-3 mini | 90.1% ±0.5 | reasoning | Apr 2025 | Official leaderboard | |
| 46 | GLM-4.5 | 90% ±0.5 | Aug 2025 | Official leaderboard | ||
| 47 | | 89.5% ±0.5 | Sep 2025 | Official leaderboard | ||
| 48 | | 88.9% ±0.6 | thinking | Sep 2025 | Official leaderboard | |
| 49 | | 88.2% ±1.1 | Jul 2024 | Official leaderboard | ||
| 50 | GPT-4o Aug 2024 | 88.2% ±1.1 | Aug 2024 | Aug 2024 | Official leaderboard | |
| 51 | | 87.4% ±0.6 | Sep 2025 | Official leaderboard | ||
| 52 | | 87.4% ±0.6 | Sep 2025 | Official leaderboard | ||
| 53 | | 84.8% ±1.3 | Jul 2024 | Official leaderboard | ||
| 54 | GPT-4.1 mini | 84.6% ±0.6 | Apr 2025 | Official leaderboard | ||
| 55 | Kimi K2 Jul 2025 | 84% ±0.7 | Jul 2025 | Official leaderboard | ||
| 56 | Grok 3 | 83.8% ±0.7 | Apr 2025 | Official leaderboard | ||
| 57 | Claude 3.5 Sonnet | 83.2% ±1.3 | Jun 2024 | Official leaderboard | ||
| 58 | gpt-oss-20b | 82.9% ±0.7 | Aug 2025 | Official leaderboard | ||
| 59 | | 82.4% ±0.7 | Sep 2025 | Official leaderboard | ||
| 60 | | 82.2% ±0.7 | Dec 2025 | Official leaderboard | ||
| 61 | | 82% ±0.7 | Mar 2025 | Official leaderboard | ||
| 62 | GPT-4 Turbo Apr 2024 | 82% ±1.3 | Apr 2024 | Official leaderboard | ||
| 63 | | 81.5% ±1.4 | Feb 2025 | Official leaderboard | ||
| 64 | | 80.5% ±0.7 | Mar 2025 | Official leaderboard | ||
| 65 | Claude Haiku 4.5 | 79.6% ±0.7 | thinking | Oct 2025 | Official leaderboard | |
| 66 | | 78.2% ±0.7 | Aug 2025 | Official leaderboard | ||
| 67 | | 77.4% ±1.4 | Sep 2024 | Official leaderboard | ||
| 68 | | 76.5% ±1.5 | Sep 2024 | Official leaderboard | ||
| 69 | | 76.2% ±1.5 | Nov 2024 | Official leaderboard | ||
| 70 | GPT-4o mini | 72.4% ±1.6 | Jul 2024 | Official leaderboard | ||
| 71 | | 69.1% ±0.8 | Mar 2025 | Official leaderboard | ||
| 72 | GPT-4.1 nano | 68.2% ±0.8 | Apr 2025 | Official leaderboard | ||
| 73 | Jamba 1.5 Large | 68.1% ±1.6 | Aug 2024 | Official leaderboard | ||
| 74 | | 62.6% ±1.7 | Jul 2024 | Official leaderboard | ||
| 75 | | 62.1% ±1.7 | Apr 2024 | Official leaderboard | ||
| 76 | GPT 3.5 | 58.5% ±1.7 | Official leaderboard | |||
| 77 | mistral-small-2402 | 57% ±1.7 | Feb 2024 | Official leaderboard | ||
| 78 | Jamba 1.5 Mini | 55.2% ±1.7 | Aug 2024 | Official leaderboard | ||
| 79 | | 53.2% ±1.7 | Dec 2023 | Official leaderboard | ||
| 80 | Jamba 1.6 Mini | 52.5% ±0.9 | Mar 2025 | Official leaderboard | ||
| 81 | | 50.9% ±0.9 | Apr 2025 | Official leaderboard | ||
| 82 | Jamba 1.6 Large | 50.7% ±0.9 | Mar 2025 | Official leaderboard | ||
| 83 | | 43.3% ±0.9 | Apr 2025 | Official leaderboard | ||
| 84 | | 2.65% ±0.6 | Aug 2024 | Official leaderboard |
| # | Model | Score ± | Setting | Released | Run on | Source |
|---|---|---|---|---|---|---|
| 1 | o1 | 96.5% ±0.6 | Dec 2024 | Official leaderboard | ||
| 2 | GPT-5.1 | 96.4% ±0.3 | Nov 2025 | Official leaderboard | ||
| 3 | | 96.4% ±0.3 | Feb 2026 | Official leaderboard | ||
| 4 | GPT-5 | 96.3% ±0.3 | Aug 2025 | Official leaderboard | ||
| 5 | GPT-5.4 | 96.1% ±0.3 | xhigh | Mar 2026 | Official leaderboard | |
| 6 | o3 | 96.1% ±0.3 | Apr 2025 | Official leaderboard | ||
| 7 | GPT-5 mini | 96.1% ±0.3 | Aug 2025 | Official leaderboard | ||
| 8 | | 96% ±0.3 | Nov 2025 | Official leaderboard | ||
| 9 | o4-mini | 96% ±0.3 | Apr 2025 | Official leaderboard | ||
| 10 | Claude Opus 4.5 | 95.9% ±0.4 | thinking | Nov 2025 | Official leaderboard | |
| 11 | | 95.8% ±0.4 | Dec 2025 | Official leaderboard | ||
| 12 | Claude Opus 4.6 | 95.4% ±0.4 | thinking | Feb 2026 | Official leaderboard | |
| 13 | | 95.2% ±0.4 | thinking | Feb 2026 | Official leaderboard | |
| 14 | o3-mini | 94.8% ±0.8 | Jan 2025 | Official leaderboard | ||
| 15 | Claude Sonnet 4.5 | 94.7% ±0.4 | thinking | Sep 2025 | Official leaderboard | |
| 16 | Grok 4.20 | 94.5% ±0.4 | reasoning | Feb 2026 | Official leaderboard | |
| 17 | Kimi K2.5 | 94.4% ±0.4 | thinking | Jan 2026 | Official leaderboard | |
| 18 | GLM-5 | 94.3% ±0.4 | thinking | Feb 2026 | Official leaderboard | |
| 19 | GPT-5.2 | 94.1% ±0.4 | Dec 2025 | Official leaderboard | ||
| 20 | | 93.9% ±0.4 | thinking | Dec 2025 | Official leaderboard | |
| 21 | GLM-4.7 | 93.7% ±0.4 | Dec 2025 | Official leaderboard | ||
| 22 | Claude Opus 4.1 | 93.6% ±0.4 | thinking | Aug 2025 | Official leaderboard | |
| 23 | GPT-5 nano | 93.3% ±0.4 | Aug 2025 | Official leaderboard | ||
| 24 | | 93.1% ±0.5 | Jun 2025 | Official leaderboard | ||
| 25 | o1-preview | 93% ±0.9 | Sep 2024 | Official leaderboard | ||
| 26 | Claude Opus 4 | 92.9% ±0.5 | non-thinking | May 2025 | Official leaderboard | |
| 27 | Claude Sonnet 4 | 92.7% ±0.5 | thinking | May 2025 | Official leaderboard | |
| 28 | Kimi K2 Thinking | 92.6% ±0.5 | Nov 2025 | Official leaderboard | ||
| 29 | | 92.5% ±0.5 | Feb 2026 | Official leaderboard | ||
| 30 | Grok 4 | 92.5% ±0.5 | Jul 2025 | Official leaderboard | ||
| 31 | Grok-2 Dec 2024 | 92.3% ±0.9 | Dec 2024 | Official leaderboard | ||
| 32 | GLM-4.6 | 92.2% ±0.5 | Sep 2025 | Official leaderboard | ||
| 33 | Grok 4.1 Fast | 92.1% ±0.5 | reasoning | Nov 2025 | Official leaderboard | |
| 34 | Grok 4 Fast | 92.1% ±0.5 | reasoning | Sep 2025 | Official leaderboard | |
| 35 | Claude Sonnet 4.6 | 92.1% ±0.5 | Feb 2026 | Official leaderboard | ||
| 36 | | 91.4% ±0.5 | non-thinking | Sep 2025 | Official leaderboard | |
| 37 | gpt-oss-120b | 91.4% ±0.5 | Aug 2025 | Official leaderboard | ||
| 38 | GPT-4.1 | 91.2% ±0.5 | Apr 2025 | Official leaderboard | ||
| 39 | | 91.2% ±0.5 | Dec 2025 | Official leaderboard | ||
| 40 | | 91% ±0.5 | thinking | Apr 2025 | Official leaderboard | |
| 41 | | 90.8% ±1 | May 2025 | Official leaderboard | ||
| 42 | | 90.6% ±0.5 | Apr 2025 | Official leaderboard | ||
| 43 | o1-mini | 90.2% | Sep 2024 | Official leaderboard | ||
| 44 | Claude 3.7 Sonnet | 90.2% ±1 | thinking | Feb 2025 | Official leaderboard | |
| 45 | Grok-3 mini | 90.1% ±0.5 | reasoning | Apr 2025 | Official leaderboard | |
| 46 | GLM-4.5 | 90% ±0.5 | Aug 2025 | Official leaderboard | ||
| 47 | | 89.5% ±0.5 | Sep 2025 | Official leaderboard | ||
| 48 | | 88.9% ±0.6 | thinking | Sep 2025 | Official leaderboard | |
| 49 | | 88.2% ±1.1 | Jul 2024 | Official leaderboard | ||
| 50 | GPT-4o Aug 2024 | 88.2% ±1.1 | Aug 2024 | Aug 2024 | Official leaderboard | |
| 51 | | 87.4% ±0.6 | Sep 2025 | Official leaderboard | ||
| 52 | | 87.4% ±0.6 | Sep 2025 | Official leaderboard | ||
| 53 | | 84.8% ±1.3 | Jul 2024 | Official leaderboard | ||
| 54 | GPT-4.1 mini | 84.6% ±0.6 | Apr 2025 | Official leaderboard | ||
| 55 | Kimi K2 Jul 2025 | 84% ±0.7 | Jul 2025 | Official leaderboard | ||
| 56 | Grok 3 | 83.8% ±0.7 | Apr 2025 | Official leaderboard | ||
| 57 | Claude 3.5 Sonnet | 83.2% ±1.3 | Jun 2024 | Official leaderboard | ||
| 58 | gpt-oss-20b | 82.9% ±0.7 | Aug 2025 | Official leaderboard | ||
| 59 | | 82.4% ±0.7 | Sep 2025 | Official leaderboard | ||
| 60 | | 82.2% ±0.7 | Dec 2025 | Official leaderboard | ||
| 61 | | 82% ±0.7 | Mar 2025 | Official leaderboard | ||
| 62 | GPT-4 Turbo Apr 2024 | 82% ±1.3 | Apr 2024 | Official leaderboard | ||
| 63 | | 81.5% ±1.4 | Feb 2025 | Official leaderboard | ||
| 64 | | 80.5% ±0.7 | Mar 2025 | Official leaderboard | ||
| 65 | Claude Haiku 4.5 | 79.6% ±0.7 | thinking | Oct 2025 | Official leaderboard | |
| 66 | | 78.2% ±0.7 | Aug 2025 | Official leaderboard | ||
| 67 | | 77.4% ±1.4 | Sep 2024 | Official leaderboard | ||
| 68 | | 76.5% ±1.5 | Sep 2024 | Official leaderboard | ||
| 69 | | 76.2% ±1.5 | Nov 2024 | Official leaderboard | ||
| 70 | GPT-4o mini | 72.4% ±1.6 | Jul 2024 | Official leaderboard | ||
| 71 | | 69.1% ±0.8 | Mar 2025 | Official leaderboard | ||
| 72 | GPT-4.1 nano | 68.2% ±0.8 | Apr 2025 | Official leaderboard | ||
| 73 | Jamba 1.5 Large | 68.1% ±1.6 | Aug 2024 | Official leaderboard | ||
| 74 | | 62.6% ±1.7 | Jul 2024 | Official leaderboard | ||
| 75 | | 62.1% ±1.7 | Apr 2024 | Official leaderboard | ||
| 76 | mistral-small-2402 | 57% ±1.7 | Feb 2024 | Official leaderboard | ||
| 77 | Jamba 1.5 Mini | 55.2% ±1.7 | Aug 2024 | Official leaderboard | ||
| 78 | | 53.2% ±1.7 | Dec 2023 | Official leaderboard | ||
| 79 | Jamba 1.6 Mini | 52.5% ±0.9 | Mar 2025 | Official leaderboard | ||
| 80 | | 50.9% ±0.9 | Apr 2025 | Official leaderboard | ||
| 81 | Jamba 1.6 Large | 50.7% ±0.9 | Mar 2025 | Official leaderboard | ||
| 82 | | 43.3% ±0.9 | Apr 2025 | Official leaderboard | ||
| 83 | | 2.65% ±0.6 | Aug 2024 | Official leaderboard |
Rankings by part
We read 6 parts from Vals AI’s leaderboard, each ranking models on one part of the test, such as Unbiased, Hispanic and Black. o1 leads 3 of them, and 2 other models lead the rest. Arrows show how far a model moved from its overall rank among the models we track.
Unbiased
82 models we track have a score here.
- 1 o1 96.8%
- 2
Gemini 3.1 Pro ↑1 96.8%
- 3 GPT-5.1 ↓1 96.8%
- 4 GPT-5 96.7%
- 5
Gemini 3 Pro ↑3 96.5%
- 6 Claude Opus 4.5 ↑4 96.4%
- 7 Claude Opus 4.6 ↑5 96.4%
- 8 GPT-5 mini ↓1 96.3%
- 9 o3 ↓3 96.3%
- 10 GPT-5.4 ↓5 96.3%
Hispanic
82 models we track have a score here.
- 1 GPT-5.1 ↑1 96.5%
- 2 o1 ↓1 96.5%
- 3 o3 ↑3 96.2%
- 4 GPT-5 96%
- 5 GPT-5 mini ↑2 96%
- 6
Gemini 3.1 Pro ↓3 96%
- 7 o4-mini ↑2 95.9%
- 8 GPT-5.4 ↓3 95.7%
- 9
Gemini 3 Pro ↓1 95.5%
- 10
Gemini 3 Flash ↑1 95.4%
Black
82 models we track have a score here.
- 1
Gemini 3.1 Pro ↑2 96.8%
- 2 o1 ↓1 96.6%
- 3 GPT-5 ↑1 96.5%
- 4 GPT-5.4 ↑1 96.5%
- 5 GPT-5 mini ↑2 96.3%
- 6 GPT-5.1 ↓4 96.2%
- 7 o4-mini ↑2 96.1%
- 8 Claude Opus 4.5 ↑2 96%
- 9
Gemini 3 Pro ↓1 95.8%
- 10 o3 ↓4 95.8%
Asian
82 models we track have a score here.
- 1 o1 96.4%
- 2 o3 ↑4 96.3%
- 3 GPT-5 ↑1 96.3%
- 4
Gemini 3 Pro ↑4 96.3%
- 5
Gemini 3.1 Pro ↓2 96.3%
- 6 GPT-5 mini ↑1 96.2%
- 7 GPT-5.4 ↓2 96.2%
- 8 GPT-5.1 ↓6 96.1%
- 9 o4-mini 96%
- 10 Claude Opus 4.5 96%
White
82 models we track have a score here.
- 1 o1 96.7%
- 2 GPT-5.1 96.5%
- 3 GPT-5 ↑1 96.3%
- 4
Gemini 3 Flash ↑7 96.3%
- 5
Gemini 3.1 Pro ↓2 96.3%
- 6 GPT-5 mini ↑1 96.2%
- 7 Claude Opus 4.5 ↑3 96.2%
- 8
Gemini 3 Pro 96.2%
- 9 GPT-5.4 ↓4 96%
- 10 o4-mini ↓1 96%
Indigenous
82 models we track have a score here.
- 1 GPT-5.1 ↑1 96.3%
- 2 o1 ↓1 96.2%
- 3
Gemini 3.1 Pro 96.1%
- 4 GPT-5 96%
- 5 o3 ↑1 95.9%
- 6 o4-mini ↑3 95.9%
- 7
Gemini 3 Pro ↑1 95.9%
- 8 GPT-5.4 ↓3 95.9%
- 9
Gemini 3 Flash ↑2 95.6%
- 10 Claude Opus 4.5 95.5%
The best score, 96.5%, costs $0.05 a task
Runs on the cost frontier score higher than any run that's cheaper. Grok 4.1 Fast at reasoning comes within 4.4 points of the best for $0.0006 a task, 93 times cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.
Source: Vals AI’s leaderboard
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| o1 | 96.5% | $0.05 | |||
| GPT-5.1 | 96.4% | $0.01 | |||
| 96.4% | $0.02 | ||||
| GPT-5 | 96.3% | $0.02 | |||
| GPT-5.4 | xhigh | 96.1% | $0.04 | ||
| o3 | 96.1% | $0.005 | |||
| GPT-5 mini | 96.1% | $0.003 | |||
| 96% | $0.03 | ||||
| o4-mini | 96% | $0.003 | |||
| Claude Opus 4.5 | thinking | 95.9% | $0.15 | ||
| 95.8% | $0.01 | ||||
| Claude Opus 4.6 | thinking | 95.4% | $0.03 | ||
| thinking | 95.2% | $0.008 | |||
| o3-mini | 94.8% | $0.005 | |||
| Claude Sonnet 4.5 | thinking | 94.7% | $0.03 | ||
| Grok 4.20 | reasoning | 94.5% | $0.009 | ||
| Kimi K2.5 | thinking | 94.4% | $0.007 | ||
| GLM-5 | thinking | 94.3% | $0.008 | ||
| GPT-5.2 | 94.1% | $0.02 | |||
| thinking | 93.9% | $0.0008 | |||
| GLM-4.7 | 93.7% | $0.005 | |||
| Claude Opus 4.1 | thinking | 93.6% | $0.09 | ||
| GPT-5 nano | 93.3% | $0.001 | |||
| Claude Opus 4.5 | non-thinking | 93.2% | $0.04 | ||
| 93.1% | $0.003 | ||||
| o1-preview | 93% | $0.11 | |||
| Claude Opus 4 | non-thinking | 92.9% | $0.04 | ||
| Claude Sonnet 4 | thinking | 92.7% | $0.03 | ||
| Kimi K2 Thinking | 92.6% | $0.006 | |||
| Claude Opus 4.1 | non-thinking | 92.5% | $0.04 | ||
| 92.5% | $0.003 | ||||
| Grok 4 | 92.5% | $0.03 | |||
| Grok-2 Dec 2024 | 92.3% | $0.003 | |||
| GLM-4.6 | 92.2% | $0.004 | |||
| Grok 4.1 Fast | reasoning | 92.1% | $0.0006 | ||
| Grok 4 Fast | reasoning | 92.1% | $0.0007 | ||
| Claude Sonnet 4.6 | 92.1% | $0.04 | |||
| non-thinking | 91.4% | $0.001 | |||
| gpt-oss-120b | 91.4% | $0.0006 | |||
| GPT-4.1 | 91.2% | $0.003 | |||
| thinking | 91.2% | $0.001 | |||
| 91.2% | $0.002 | ||||
| thinking | 91% | $0.004 | |||
| 90.8% | $0.006 | ||||
| 90.6% | $0.001 | ||||
| Claude Sonnet 4 | non-thinking | 90.3% | $0.007 | ||
| o1-mini | 90.2% | ||||
| Claude 3.7 Sonnet | thinking | 90.2% | $0.02 | ||
| Grok-3 mini | reasoning | 90.1% | $0.0008 | ||
| GLM-4.5 | 90% | $0.003 | |||
| 89.5% | $0.007 | ||||
| non-thinking | 89.5% | $0.0002 | |||
| thinking | 88.9% | $0.0002 | |||
| Grok-3 mini | reasoning | 88.7% | $0.0006 | ||
| 88.2% | $0.003 | ||||
| GPT-4o Aug 2024 | 88.2% | $0.003 | |||
| 87.4% | $0.003 | ||||
| 87.4% | $0.003 | ||||
| non-thinking | 86.7% | $0.001 | |||
| 84.8% | $0.0006 | ||||
| GPT-4.1 mini | 84.6% | $0.0005 | |||
| Kimi K2 Jul 2025 | 84% | $0.001 | |||
| Grok 3 | 83.8% | $0.006 | |||
| Claude 3.5 Sonnet | 83.2% | $0.006 | |||
| gpt-oss-20b | 82.9% | $0.0003 | |||
| 82.4% | $0.002 | ||||
| 82.2% | $0.0008 | ||||
| 82% | $0.0006 | ||||
| GPT-4 Turbo Apr 2024 | 82% | $0.01 | |||
| 81.5% | $0.0001 | ||||
| 80.9% | $0.0006 | ||||
| 80.5% | $0.003 | ||||
| non-thinking | 80.3% | $0.0003 | |||
| Claude Haiku 4.5 | thinking | 79.6% | $0.01 | ||
| 78.2% | $0.0006 | ||||
| 77.4% | $0.0008 | ||||
| 76.5% | $0.003 | ||||
| 76.2% | $0.002 | ||||
| Grok 4.1 Fast | non-reasoning | 76% | $0.0002 | ||
| Grok 4 Fast | non-reasoning | 75.4% | $0.0005 | ||
| GPT-4o mini | 72.4% | $0.0002 | |||
| 69.1% | $0.0001 | ||||
| GPT-4.1 nano | 68.2% | $0.0002 | |||
| Jamba 1.5 Large | 68.1% | $0.002 | |||
| 62.6% | $0.0001 | ||||
| 62.1% | $0.0009 | ||||
| mistral-small-2402 | 57% | $0.0002 | |||
| Jamba 1.5 Mini | 55.2% | $0.0001 | |||
| 53.2% | $0.0004 | ||||
| Jamba 1.6 Mini | 52.5% | $0.0002 | |||
| 50.9% | $0.0003 | ||||
| Jamba 1.6 Large | 50.7% | $0.003 | |||
| 43.3% | $0.0007 | ||||
| 2.65% | $0.003 |
The best score, 96.5%, costs $0.05 a task
Runs on the cost frontier score higher than any run that's cheaper. Grok 4.1 Fast at reasoning comes within 4.4 points of the best for $0.0006 a task, 93 times cheaper. The costs are the leaderboard's own estimates, so compare them only within this benchmark.
Source: Vals AI’s leaderboard
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| o1 | 96.5% | $0.05 | |||
| GPT-5.1 | 96.4% | $0.01 | |||
| 96.4% | $0.02 | ||||
| GPT-5 | 96.3% | $0.02 | |||
| GPT-5.4 | xhigh | 96.1% | $0.04 | ||
| o3 | 96.1% | $0.005 | |||
| GPT-5 mini | 96.1% | $0.003 | |||
| 96% | $0.03 | ||||
| o4-mini | 96% | $0.003 | |||
| Claude Opus 4.5 | thinking | 95.9% | $0.15 | ||
| 95.8% | $0.01 | ||||
| Claude Opus 4.6 | thinking | 95.4% | $0.03 | ||
| thinking | 95.2% | $0.008 | |||
| o3-mini | 94.8% | $0.005 | |||
| Claude Sonnet 4.5 | thinking | 94.7% | $0.03 | ||
| Grok 4.20 | reasoning | 94.5% | $0.009 | ||
| Kimi K2.5 | thinking | 94.4% | $0.007 | ||
| GLM-5 | thinking | 94.3% | $0.008 | ||
| GPT-5.2 | 94.1% | $0.02 | |||
| thinking | 93.9% | $0.0008 | |||
| GLM-4.7 | 93.7% | $0.005 | |||
| Claude Opus 4.1 | thinking | 93.6% | $0.09 | ||
| GPT-5 nano | 93.3% | $0.001 | |||
| Claude Opus 4.5 | non-thinking | 93.2% | $0.04 | ||
| 93.1% | $0.003 | ||||
| o1-preview | 93% | $0.11 | |||
| Claude Opus 4 | non-thinking | 92.9% | $0.04 | ||
| Claude Sonnet 4 | thinking | 92.7% | $0.03 | ||
| Kimi K2 Thinking | 92.6% | $0.006 | |||
| Claude Opus 4.1 | non-thinking | 92.5% | $0.04 | ||
| 92.5% | $0.003 | ||||
| Grok 4 | 92.5% | $0.03 | |||
| Grok-2 Dec 2024 | 92.3% | $0.003 | |||
| GLM-4.6 | 92.2% | $0.004 | |||
| Grok 4.1 Fast | reasoning | 92.1% | $0.0006 | ||
| Grok 4 Fast | reasoning | 92.1% | $0.0007 | ||
| Claude Sonnet 4.6 | 92.1% | $0.04 | |||
| non-thinking | 91.4% | $0.001 | |||
| gpt-oss-120b | 91.4% | $0.0006 | |||
| GPT-4.1 | 91.2% | $0.003 | |||
| thinking | 91.2% | $0.001 | |||
| 91.2% | $0.002 | ||||
| thinking | 91% | $0.004 | |||
| 90.8% | $0.006 | ||||
| 90.6% | $0.001 | ||||
| Claude Sonnet 4 | non-thinking | 90.3% | $0.007 | ||
| o1-mini | 90.2% | ||||
| Claude 3.7 Sonnet | thinking | 90.2% | $0.02 | ||
| Grok-3 mini | reasoning | 90.1% | $0.0008 | ||
| GLM-4.5 | 90% | $0.003 | |||
| 89.5% | $0.007 | ||||
| non-thinking | 89.5% | $0.0002 | |||
| thinking | 88.9% | $0.0002 | |||
| Grok-3 mini | reasoning | 88.7% | $0.0006 | ||
| 88.2% | $0.003 | ||||
| GPT-4o Aug 2024 | 88.2% | $0.003 | |||
| 87.4% | $0.003 | ||||
| 87.4% | $0.003 | ||||
| non-thinking | 86.7% | $0.001 | |||
| 84.8% | $0.0006 | ||||
| GPT-4.1 mini | 84.6% | $0.0005 | |||
| Kimi K2 Jul 2025 | 84% | $0.001 | |||
| Grok 3 | 83.8% | $0.006 | |||
| Claude 3.5 Sonnet | 83.2% | $0.006 | |||
| gpt-oss-20b | 82.9% | $0.0003 | |||
| 82.4% | $0.002 | ||||
| 82.2% | $0.0008 | ||||
| 82% | $0.0006 | ||||
| GPT-4 Turbo Apr 2024 | 82% | $0.01 | |||
| 81.5% | $0.0001 | ||||
| 80.9% | $0.0006 | ||||
| 80.5% | $0.003 | ||||
| non-thinking | 80.3% | $0.0003 | |||
| Claude Haiku 4.5 | thinking | 79.6% | $0.01 | ||
| 78.2% | $0.0006 | ||||
| 77.4% | $0.0008 | ||||
| 76.5% | $0.003 | ||||
| 76.2% | $0.002 | ||||
| Grok 4.1 Fast | non-reasoning | 76% | $0.0002 | ||
| Grok 4 Fast | non-reasoning | 75.4% | $0.0005 | ||
| GPT-4o mini | 72.4% | $0.0002 | |||
| 69.1% | $0.0001 | ||||
| GPT-4.1 nano | 68.2% | $0.0002 | |||
| Jamba 1.5 Large | 68.1% | $0.002 | |||
| 62.6% | $0.0001 | ||||
| 62.1% | $0.0009 | ||||
| mistral-small-2402 | 57% | $0.0002 | |||
| Jamba 1.5 Mini | 55.2% | $0.0001 | |||
| 53.2% | $0.0004 | ||||
| Jamba 1.6 Mini | 52.5% | $0.0002 | |||
| 50.9% | $0.0003 | ||||
| Jamba 1.6 Large | 50.7% | $0.003 | |||
| 43.3% | $0.0007 | ||||
| 2.65% | $0.003 |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Vals AI’s leaderboard
Llama 3.1-405B on the official leaderboard passed 85% in July 2024
Models released before it came out already scored that high. All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Vals AI’s leaderboard
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.
More in Law, finance and medicine
- MedXpertQA-MM In use Models answer expert-level medical exam questions that combine images with clinical records and test results. Scoring is accuracy.
Qwen 3.8 Max 80.4%
- HealthBench Professional In use 525 physician-authored tasks from real clinician chats (care consult, documentation, research) with physician-adjudicated rubrics. Claude Sonnet 5.5 69.2%
- HealthBench In use 5,000 realistic health conversations graded against physician-written rubrics. A length-adjusted variant penalizes verbosity. GPT-5 67.2%
- FinSearchComp In use Agents search the web to fetch live financial data, look up history, and run multi-step investigations, across 635 expert-written questions. Scored by answer accuracy.
Ling-3.1-flash 81.7%
- KorMedMCQA In use Models answer 7,469 multiple-choice questions from Korean licensing exams for doctors, nurses, pharmacists and dentists, held from 2012 to 2024. o1-preview 92.7%
- Harvey LAB-AA v1.1 In use Artificial Analysis's run of Harvey's Legal Agent Benchmark on Harvey's newer private set of 120 tasks across 24 practice areas. A deliverable with a material hallucination scores zero, and the headline is the share of judges who find every rubric criterion passed. GPT-6.2 Astra 15%
Data audit
7 of 8 checks pass, 1 warning. Look first at the Task access check. We don’t know whether the tasks are public. Every benchmark
Catalog record 1 warning 1 pass
- Warning Task access We don’t know whether the tasks are public.
- Pass Confidence Checked against a primary source.
Scores 2 pass 3 notes
- Pass Independent runs Scores from the official leaderboard (84 entries).
- Pass Models scored 83 models we track, plus 1 other entry on its leaderboard.
- Note Newest score The newest came 8 months ago, in March 2026, for GPT-5.4. That’s expected for a saturated test.
- Note Frontier models None of the top 10 has a score, which is fine for a retired test. The strongest tested is GPT-5.4 (#28).
- Note Headroom Saturated. The best is 96.5%, past 85% since July 2024.
Leaderboard import 3 pass
- Pass Importer We read the board on Vals AI: 94 rows for models we track and 1 other entry at the last read.
- Pass Last read November 20th, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.