benchgap
Knowledge & reasoning

HealthBench Professional (raw) leaderboard

As of 2026-11-21, the highest measured score on HealthBench Professional (raw) is 77.1% by Claude Opus 5.5. 115 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#ModelScoreSource
1Claude Opus 5.577.1%measured
2Claude Sonnet 5.577.1%measured
3Claude Fable 5.174.3%estimated ± 4.5 pp, high confidence
4Muse Spark74.3%estimated ± 1.7 pp, low confidence
5Claude Opus 573.4%measured
6Claude Fable 571.7%estimated ± 4.5 pp, high confidence
7GPT-5.471.7%estimated ± 1.7 pp, low confidence
8Muse Wonder Full70.3%estimated ± 4.5 pp, high confidence
9Kimi K3.1 Pro69.5%estimated ± 4.5 pp, high confidence
10GPT-6 Astra68.2%measured
11Muse Wonder67.6%estimated ± 4.5 pp, high confidence
12GPT-6.1 Sol67.2%measured
13MiMo-V2.6-Pro67.1%estimated ± 4.5 pp, high confidence
14Gemini 4 Argon66.6%estimated ± 2.3 pp, medium confidence
15Claude Opus 4.866.5%estimated ± 4.5 pp, high confidence
16Muse Spark 1.366.5%estimated ± 4.5 pp, high confidence
17Gemini 4 Carbon66.5%estimated ± 2.3 pp, medium confidence
18Claude Fable 5.266.4%estimated ± 2.3 pp, medium confidence
19GPT-6.2 Astra66.0%measured
20MiniMax M3.1 Flash66.0%estimated ± 4.5 pp, high confidence
21Gemini 3.7 Flash65.9%estimated ± 4.5 pp, high confidence
22GPT-6.2 Sol65.5%measured
23Gemini 3.1 Pro65.2%estimated ± 4.5 pp, high confidence
24Kimi K365.1%estimated ± 4.5 pp, high confidence
25Step 5 Preview64.8%estimated ± 4.5 pp, high confidence
26Muse Spark 1.164.5%estimated ± 4.5 pp, high confidence
27GPT-5.564.2%estimated ± 4.5 pp, high confidence
28GPT-6.1 Terra64.2%measured
29GPT-5.6 Sol64.1%estimated ± 1.7 pp, medium confidence
30Muse Spark 1.264.0%estimated ± 4.5 pp, high confidence
31GPT-5.6 Terra63.6%estimated ± 1.7 pp, medium confidence
32GPT-5.6 Luna62.8%estimated ± 1.7 pp, medium confidence
33Muse Spark Open62.6%estimated ± 4.5 pp, high confidence
34Gemini 3.8 Flash62.2%estimated ± 2.3 pp, medium confidence
35Qwen3.8 Max Preview62.0%estimated ± 4.5 pp, high confidence
36Grok 4.661.8%estimated ± 4.5 pp, high confidence
37Gemini 3.5 Flash61.6%estimated ± 4.5 pp, high confidence
38Grok 4.561.6%estimated ± 4.5 pp, high confidence
39GPT-5.3 Codex61.5%estimated ± 4.5 pp, high confidence
40Claude Opus 4.761.3%estimated ± 4.5 pp, high confidence
41GLM-5.361.3%estimated ± 4.5 pp, high confidence
42GPT-6 Luna61.2%measured
43Claude Sonnet 560.4%estimated ± 4.5 pp, high confidence
44GLM-5.260.3%estimated ± 4.5 pp, high confidence
45Grok 4.760.2%estimated ± 2.3 pp, medium confidence
46DeepSeek V4 Pro60.2%estimated ± 4.5 pp, high confidence
47Gemini 3.6 Flash60.0%estimated ± 4.5 pp, high confidence
48Claude Haiku 5.559.9%estimated ± 2.3 pp, medium confidence
49Inkling 2 Small59.9%estimated ± 2.3 pp, medium confidence
50Qwen3.7 Max59.8%estimated ± 4.5 pp, high confidence
51GPT-6 Sol59.5%measured
52Claude Opus 4.6 (Adaptive)59.2%estimated ± 4.5 pp, high confidence
53Gemini 3 Pro59.1%estimated ± 4.5 pp, high confidence
54Ling 3.1 Flash58.8%estimated ± 4.5 pp, high confidence
55DeepSeek V4.1 Flash 101758.6%estimated ± 4.5 pp, high confidence
56MiniMax M358.4%estimated ± 4.5 pp, high confidence
57DeepSeek V4 Flash 073158.1%estimated ± 4.5 pp, high confidence
58Qwen3.8-Flash-Next57.6%estimated ± 4.5 pp, medium confidence
59GPT-5.257.3%estimated ± 4.5 pp, medium confidence
60Kimi K2.657.1%estimated ± 4.5 pp, medium confidence
61Grok 4.356.8%estimated ± 4.5 pp, medium confidence
62GPT-5.2-Codex55.5%estimated ± 4.5 pp, medium confidence
63MiMo-V2.5-Pro55.5%estimated ± 4.5 pp, medium confidence
64Qwen3.7 Plus55.4%estimated ± 4.5 pp, medium confidence
65MiMo-V2.6-Flash54.9%estimated ± 4.5 pp, medium confidence
66Kimi K2.7 Code54.8%estimated ± 4.5 pp, medium confidence
67Beam 1.0 501B-A23B54.6%estimated ± 4.5 pp, medium confidence
68Apodex 1.154.0%estimated ± 4.5 pp, medium confidence
69Apodex 1.1 Mini54.0%estimated ± 4.5 pp, medium confidence
70Qwen3.8-27B53.8%estimated ± 4.5 pp, medium confidence
71Hy353.5%estimated ± 4.5 pp, medium confidence
72Hy3 Preview53.5%estimated ± 4.5 pp, medium confidence
73Inkling-Small53.3%estimated ± 4.5 pp, medium confidence
74Inkling51.9%estimated ± 4.5 pp, medium confidence
75Qwen 3.6 Max (preview)50.9%estimated ± 4.5 pp, medium confidence
76Kimi K2.550.8%estimated ± 4.5 pp, medium confidence
77Nemotron 3.5 Super VL 120B A12B50.8%estimated ± 4.5 pp, medium confidence
78Nemotron 3.5 Super VL 120B A12B NVFP450.8%estimated ± 4.5 pp, medium confidence
79MiMo-V2-Pro50.5%estimated ± 4.5 pp, medium confidence
80Claude Opus 4.5 Thinking50.2%estimated ± 4.5 pp, medium confidence
81GLM-5.150.2%estimated ± 4.5 pp, medium confidence
82Solar Pro 449.9%estimated ± 4.5 pp, medium confidence
83A.X K249.7%estimated ± 4.5 pp, medium confidence
84MiniMax M2.749.7%estimated ± 4.5 pp, medium confidence
85GLM-549.4%estimated ± 4.5 pp, medium confidence
86Qwen3.5 397B49.1%estimated ± 4.5 pp, medium confidence
87GPT-548.6%estimated ± 4.5 pp, medium confidence
88GPT-5.148.6%estimated ± 4.5 pp, medium confidence
89Nemotron 3 Ultra48.5%estimated ± 4.5 pp, medium confidence
90GPT-5.4 nano48.4%estimated ± 4.5 pp, medium confidence
91GPT-5.4 mini48.2%estimated ± 4.5 pp, medium confidence
92GLM-5-Turbo47.9%estimated ± 4.5 pp, medium confidence
93Qwen3.6 Plus47.9%estimated ± 4.5 pp, medium confidence
94GLM-4.747.5%estimated ± 4.5 pp, medium confidence
95Grok 446.8%estimated ± 4.5 pp, medium confidence
96GPT-5.1-Codex45.8%estimated ± 4.5 pp, medium confidence
97GPT-5.1-Codex-Max45.8%estimated ± 4.5 pp, medium confidence
98Qwen3.5-122B-A10B45.3%estimated ± 4.5 pp, medium confidence
99Nemotron 3.5 Bolt 74B-A9B44.8%estimated ± 4.5 pp, medium confidence
100Qwen3.5-27B43.9%estimated ± 4.5 pp, medium confidence
101Ling 3.0 Flash43.7%estimated ± 4.5 pp, medium confidence
102Gemma 4 31B43.6%estimated ± 4.5 pp, medium confidence
103Qwen3.6-27B43.1%estimated ± 4.5 pp, medium confidence
104Gemini 2.5 Pro42.5%estimated ± 4.5 pp, medium confidence
105Qwen3.6-35B-A3B42.1%estimated ± 4.5 pp, medium confidence
106MiMo-V2-Omni42.0%estimated ± 4.5 pp, medium confidence
107Ling 3.0 Flash VL41.9%estimated ± 4.5 pp, medium confidence
108Muse Glimmer 30B41.9%estimated ± 4.5 pp, medium confidence
109Step 3.7 Flash41.3%estimated ± 4.5 pp, medium confidence
110Qwen3.5-35B-A3B40.8%estimated ± 4.5 pp, medium confidence
111o339.9%estimated ± 4.5 pp, medium confidence
112GPT-OSS 120B39.3%estimated ± 4.5 pp, medium confidence
113Gemma 4 26B A4B39.0%estimated ± 4.5 pp, medium confidence
114Grok 4.1 Fast (Reasoning)39.0%estimated ± 4.5 pp, medium confidence
115Claude Opus 4.638.7%estimated ± 4.5 pp, medium confidence
116Grok 4 Fast (Reasoning)38.7%estimated ± 4.5 pp, medium confidence
117Gemini 3.5 Flash-Lite38.4%estimated ± 4.5 pp, medium confidence
118Quasar 438B38.3%estimated ± 4.5 pp, medium confidence
119K-EXAONE 2.038.2%estimated ± 4.5 pp, medium confidence
120GLM-5V-Turbo36.5%estimated ± 4.5 pp, medium confidence
121o4-mini35.8%estimated ± 4.5 pp, medium confidence
122DeepSeek-R135.0%estimated ± 4.5 pp, medium confidence
123Trinity-Large-Preview35.0%estimated ± 4.5 pp, medium confidence
124Trinity-Large-Thinking35.0%estimated ± 4.5 pp, medium confidence
125Gemma 4 12B34.9%estimated ± 4.5 pp, medium confidence

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General