benchgap
Knowledge & reasoning

HealthBench Professional (length-adjusted, Artificial Analysis) leaderboard

As of 2026-11-18, the highest measured score on HealthBench Professional (length-adjusted, Artificial Analysis) is 67.7% by Claude Fable 5.2. 171 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#ModelScoreSource
1Claude Fable 5.267.7%measured
2GPT-6 Astra66.0%measured
3GPT-6.2 Astra64.8%estimated ± 4.3 pp, medium confidence
4Gemini 4 Carbon64.1%estimated ± 4.3 pp, medium confidence
5GPT-5.6 Sol63.8%estimated ± 4.3 pp, medium confidence
6GPT-6.2 Sol63.3%estimated ± 4.3 pp, medium confidence
7Claude Sonnet 5.563.0%measured
8GPT-5.6 Terra62.6%estimated ± 4.3 pp, medium confidence
9GPT-6.1 Sol62.5%estimated ± 4.3 pp, medium confidence
10Gemini 4 Hydrogen62.3%estimated ± 4.3 pp, medium confidence
11DeepSeek V4 Pro62.1%estimated ± 4.3 pp, medium confidence
12GPT-6.1 Terra61.5%estimated ± 4.3 pp, medium confidence
13GPT-5.6 Luna61.5%estimated ± 4.3 pp, medium confidence
14Claude Opus 561.3%estimated ± 4.3 pp, medium confidence
15Gemini 4 Argon61.2%estimated ± 4.3 pp, medium confidence
16Gemini 3.5 Flash61.1%estimated ± 4.3 pp, medium confidence
17Muse Wonder Full61.1%estimated ± 4.3 pp, medium confidence
18Claude Fable 561.0%estimated ± 4.3 pp, medium confidence
19Muse Spark 1.360.5%estimated ± 4.3 pp, medium confidence
20GPT-6 Sol60.4%estimated ± 4.3 pp, medium confidence
21Claude Mythos 560.2%estimated ± 4.4 pp, medium confidence
22Kimi K3.1 Pro60.0%estimated ± 4.3 pp, medium confidence
23MiMo-V2.6-Pro60.0%estimated ± 4.3 pp, medium confidence
24Claude Opus 5.559.9%measured
25Grok 4.659.3%estimated ± 4.3 pp, medium confidence
26Kimi K359.1%estimated ± 4.3 pp, medium confidence
27Muse Wonder59.1%estimated ± 4.3 pp, medium confidence
28Qwen4 Max58.9%estimated ± 4.3 pp, medium confidence
29Claude Opus 4.858.6%estimated ± 4.3 pp, medium confidence
30Sakana Fugu-Ultra58.5%estimated ± 4.4 pp, medium confidence
31Claude Haiku 5.558.4%measured
32Hy3 Preview58.4%estimated ± 4.3 pp, medium confidence
33Pareto 26.958.4%estimated ± 4.4 pp, medium confidence
34Claude Opus 4.758.2%estimated ± 4.3 pp, medium confidence
35Sakana Fugu58.2%estimated ± 4.4 pp, medium confidence
36ERNIE 5.5 Flash Preview58.1%estimated ± 4.4 pp, medium confidence
37Gemini 3.1 Pro58.1%estimated ± 4.4 pp, low confidence
38Ornith-1.5-397B58.0%estimated ± 4.4 pp, low confidence
39Qwen3.8 Max57.9%estimated ± 4.4 pp, low confidence
40Hy4 preview57.9%estimated ± 4.4 pp, low confidence
41GPT-5.5 Pro57.9%estimated ± 4.4 pp, low confidence
42Muse Spark 1.257.9%estimated ± 4.3 pp, medium confidence
43GPT-5.4 Pro57.9%estimated ± 4.4 pp, low confidence
44DeepSeek V4.1 Flash 101757.9%estimated ± 4.3 pp, medium confidence
45Claude Fable 5.157.8%measured
46Gemini 3.7 Flash57.7%estimated ± 4.3 pp, medium confidence
47GPT-5.457.7%estimated ± 4.3 pp, medium confidence
48Grok 4.557.7%estimated ± 4.3 pp, medium confidence
49Beam 1.0 501B-A23B FP857.6%estimated ± 4.4 pp, low confidence
50Beam 1.0 501B-A23B NVFP457.6%estimated ± 4.4 pp, low confidence
51Grok 4.2057.5%estimated ± 4.4 pp, low confidence
52GPT-5.557.5%estimated ± 4.3 pp, medium confidence
53Sage-A1.557.5%estimated ± 7.3 pp, low confidence
54Solar Open 257.5%estimated ± 4.4 pp, low confidence
55Nemotron 3.5 Bolt 74B-A9B BF1657.5%estimated ± 4.4 pp, low confidence
56Ornith-1.5-35B-A3B57.5%estimated ± 4.4 pp, low confidence
57Nemotron 3.5 Bolt 74B-A9B57.5%estimated ± 4.4 pp, low confidence
58Ornith-1.5-9B57.5%estimated ± 4.4 pp, low confidence
59Gemma 4 31B57.5%estimated ± 4.4 pp, low confidence
60Claude Sonnet 557.5%estimated ± 4.3 pp, medium confidence
61Grok 4.357.3%estimated ± 4.3 pp, medium confidence
62Muse Ember56.8%estimated ± 4.3 pp, low confidence
63Grok 4.756.6%estimated ± 4.3 pp, low confidence
64Ling 3.1 Flash56.5%estimated ± 7.3 pp, low confidence
65DeepSeek V4 Flash 073156.3%estimated ± 4.3 pp, low confidence
66Gemini 3.6 Flash56.2%estimated ± 4.3 pp, low confidence
67Muse Spark 1.156.1%estimated ± 4.3 pp, low confidence
68GLM-5.256.1%estimated ± 4.3 pp, low confidence
69Qwen3.8 Max Preview56.1%estimated ± 4.3 pp, low confidence
70GPT-5.3 Codex56.0%estimated ± 7.3 pp, low confidence
71Claude Opus 4.6 (Adaptive)55.6%estimated ± 4.3 pp, low confidence
72GLM-5.355.5%estimated ± 4.3 pp, low confidence
73Muse Spark55.4%estimated ± 4.3 pp, low confidence
74Gemini 3.8 Flash55.2%estimated ± 4.3 pp, low confidence
75Inkling 2 Small55.1%estimated ± 4.3 pp, low confidence
76GPT-5.255.1%estimated ± 4.3 pp, low confidence
77Qwen3.7 Max54.8%estimated ± 4.3 pp, low confidence
78Claude Opus 4.5 Thinking54.7%estimated ± 4.3 pp, low confidence
79MiniMax M3.1 Flash54.7%estimated ± 4.3 pp, low confidence
80MiMo-V2-Pro54.5%estimated ± 4.3 pp, low confidence
81Qwen3.8-Flash-Next54.5%estimated ± 4.3 pp, low confidence
82Step 5 Preview54.5%estimated ± 4.3 pp, low confidence
83GPT-5.2-Codex54.5%estimated ± 4.3 pp, low confidence
84Qwen 3.6 Max (preview)54.5%estimated ± 4.3 pp, low confidence
85Gemini 3 Pro54.3%estimated ± 4.3 pp, low confidence
86GLM-554.3%estimated ± 4.3 pp, low confidence
87GPT-6 Luna54.3%estimated ± 4.3 pp, low confidence
88Qwen3.6 Plus54.0%estimated ± 4.3 pp, low confidence
89Kimi K2.654.0%estimated ± 4.3 pp, low confidence
90MiMo-V2.6-Flash54.0%estimated ± 4.3 pp, low confidence
91GLM-5-Turbo53.9%estimated ± 4.3 pp, low confidence
92Claude Opus 4.653.8%estimated ± 4.3 pp, low confidence
93Mistral Large 453.8%estimated ± 4.3 pp, low confidence
94GLM-5.153.8%estimated ± 4.3 pp, low confidence
95MiMo-V2.5-Pro53.7%estimated ± 4.3 pp, low confidence
96Muse Spark Open53.4%estimated ± 4.3 pp, low confidence
97GPT-5.153.3%estimated ± 4.3 pp, low confidence
98Claude Sonnet 4.653.3%estimated ± 4.3 pp, low confidence
99GPT-5.4 mini53.1%estimated ± 4.3 pp, low confidence
100MiMo-V2-Omni53.1%estimated ± 4.3 pp, low confidence
101GPT-5.1-Codex53.0%estimated ± 4.3 pp, low confidence
102GPT-5.1-Codex-Max53.0%estimated ± 4.3 pp, low confidence
103Claude Opus 4.553.0%estimated ± 4.3 pp, low confidence
104GLM-5V-Turbo53.0%estimated ± 4.3 pp, low confidence
105Kimi K2.553.0%estimated ± 4.3 pp, low confidence
106GPT-552.8%estimated ± 4.3 pp, low confidence
107Qwen3.5-27B52.8%estimated ± 4.3 pp, low confidence
108MiniMax M2.752.7%estimated ± 4.3 pp, low confidence
109Command A+52.7%estimated ± 4.3 pp, low confidence
110Grok 452.7%estimated ± 4.3 pp, low confidence
111GLM-4.752.6%estimated ± 4.3 pp, low confidence
112Kimi K2.7 Code52.5%estimated ± 7.3 pp, low confidence
113o3-pro52.5%estimated ± 4.3 pp, low confidence
114Qwen3.6-27B52.3%estimated ± 4.3 pp, low confidence
115Quasar 438B52.2%estimated ± 7.3 pp, low confidence
116GPT-5.4 nano52.1%estimated ± 4.3 pp, low confidence
117Claude 4 Opus Thinking52.1%estimated ± 4.3 pp, low confidence
118Qwen3.7 Plus52.1%estimated ± 7.3 pp, low confidence
119Grok 4.1 Fast (Reasoning)52.0%estimated ± 4.3 pp, low confidence
120Ling 3.0 Flash51.9%estimated ± 4.3 pp, low confidence
121Qwen3.8-27B51.9%estimated ± 4.3 pp, low confidence
122Step 3.7 Flash51.7%estimated ± 4.3 pp, low confidence
123Qwen3.5-35B-A3B51.7%estimated ± 4.3 pp, low confidence
124MiniMax M351.6%estimated ± 4.3 pp, low confidence
125Claude 4.1 Opus51.5%estimated ± 4.3 pp, low confidence
126Grok 4 Fast (Reasoning)51.3%estimated ± 4.3 pp, low confidence
127Gemini 3 Flash51.3%estimated ± 4.3 pp, low confidence
128Beam 1.0 501B-A23B51.2%estimated ± 4.3 pp, low confidence
129Claude 3.7 Sonnet Thinking51.2%estimated ± 4.3 pp, low confidence
130o4-mini50.9%estimated ± 4.3 pp, low confidence
131Claude 4 Opus50.9%estimated ± 4.3 pp, low confidence
132Claude 4 Sonnet50.9%estimated ± 4.3 pp, low confidence
133Gemini 2.5 Pro50.7%estimated ± 4.3 pp, low confidence
134MiMo-V2-Flash50.7%estimated ± 4.3 pp, low confidence
135DeepSeek V3.250.7%estimated ± 4.3 pp, low confidence
136Qwen3 Max50.6%estimated ± 4.3 pp, low confidence
137Claude 3.7 Sonnet50.4%estimated ± 4.3 pp, low confidence
138o150.4%estimated ± 4.3 pp, low confidence
139GLM-4.650.3%estimated ± 4.3 pp, low confidence
140K-Exaone50.2%estimated ± 4.3 pp, low confidence
141Solar Pro 450.1%estimated ± 4.3 pp, low confidence
142Grok Code Fast 150.1%estimated ± 4.3 pp, low confidence
143Ling 2.6 Flash50.1%estimated ± 4.3 pp, low confidence
144DeepSeek V3.150.0%estimated ± 4.3 pp, low confidence
145Apodex 1.150.0%estimated ± 4.3 pp, low confidence
146Apodex 1.1 Mini50.0%estimated ± 4.3 pp, low confidence
147Nemotron 3.5 Super VL 120B A12B49.9%estimated ± 4.3 pp, low confidence
148Nemotron 3.5 Super VL 120B A12B NVFP449.9%estimated ± 4.3 pp, low confidence
149Hy349.9%estimated ± 7.3 pp, low confidence
150DeepSeek-R149.8%estimated ± 4.3 pp, low confidence
151Inkling-Small49.7%estimated ± 4.3 pp, low confidence
152Kimi K249.7%estimated ± 4.3 pp, low confidence
153GPT-4.149.7%estimated ± 4.3 pp, low confidence
154Inkling49.6%estimated ± 4.3 pp, low confidence
155o3-mini49.6%estimated ± 4.3 pp, low confidence
156o1-pro49.6%estimated ± 4.3 pp, low confidence
157Nemotron 3 Ultra49.4%estimated ± 4.3 pp, low confidence
158Gemini 3.5 Flash-Lite49.3%estimated ± 4.3 pp, low confidence
159o1-preview49.3%estimated ± 4.3 pp, low confidence
160Grok 4.1 Fast49.2%estimated ± 4.3 pp, low confidence
161GLM-4.5-Air49.2%estimated ± 4.3 pp, low confidence
162GPT-4.1 mini48.9%estimated ± 4.3 pp, low confidence
163Gemini 2.5 Flash48.8%estimated ± 4.3 pp, low confidence
164o1-mini48.8%estimated ± 4.3 pp, low confidence
165DeepSeek V3 032448.8%estimated ± 4.3 pp, low confidence
166GPT-4.548.7%estimated ± 4.3 pp, low confidence
167QwQ-32B48.7%estimated ± 4.3 pp, low confidence
168Muse Glimmer 30B48.6%estimated ± 7.3 pp, low confidence
169Mistral Large 348.6%estimated ± 4.3 pp, low confidence
170Magistral Medium48.5%estimated ± 4.3 pp, low confidence
171Mistral Medium 348.5%estimated ± 4.3 pp, low confidence
172A.X K248.5%estimated ± 4.3 pp, low confidence
173Claude 3.5 Haiku48.5%estimated ± 4.3 pp, low confidence
174Gemini 2.0 Flash48.5%estimated ± 4.3 pp, low confidence
175Claude 3 Opus48.4%estimated ± 4.3 pp, low confidence
176DeepSeek V348.4%estimated ± 4.3 pp, low confidence
177GPT-4o48.4%estimated ± 4.3 pp, low confidence

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General