Knowledge & reasoning

HealthBench Professional (raw) leaderboard

Name: benchgap: LLM benchmark scores, measured and estimated
Creator: benchgap

As of 2026-10-07, the highest measured score on HealthBench Professional (raw) is 77.1% by Claude Opus 5.5. 193 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Claude Opus 5.5	77.1%	measured
2	Claude Sonnet 5.5	77.1%	measured
3	Claude Fable 5.1	74.3%	estimated ± 5.0 pp, medium confidence
4	Claude Opus 5	73.4%	measured
5	Gemini 4 Argon	72.9%	estimated ± 5.0 pp, medium confidence
6	Claude Fable 5	71.7%	estimated ± 5.0 pp, medium confidence
7	GPT-6 Astra	68.2%	measured
8	GPT-5.6 Sol	67.2%	estimated ± 5.0 pp, medium confidence
9	GPT-6.1 Sol	67.2%	measured
10	MiMo-V2.6-Pro	67.1%	estimated ± 5.0 pp, medium confidence
11	Claude Opus 4.8	66.6%	estimated ± 5.0 pp, medium confidence
12	Muse Spark 1.3	66.6%	estimated ± 5.0 pp, medium confidence
13	Gemini 3.7 Flash	66.0%	estimated ± 5.0 pp, medium confidence
14	Gemini 3.8 Flash	65.9%	estimated ± 5.0 pp, medium confidence
15	Gemini 3.1 Pro	65.2%	estimated ± 5.0 pp, medium confidence
16	Kimi K3	65.2%	estimated ± 5.0 pp, medium confidence
17	Step 5 Preview	64.8%	estimated ± 5.0 pp, medium confidence
18	Muse Spark 1.1	64.6%	estimated ± 5.0 pp, medium confidence
19	GPT-5.5	64.3%	estimated ± 5.0 pp, medium confidence
20	Muse Spark 1.2	64.0%	estimated ± 5.0 pp, medium confidence
21	Claude Haiku 5.5	63.1%	estimated ± 5.0 pp, medium confidence
22	GPT-5.4	62.6%	estimated ± 5.0 pp, medium confidence
23	Grok 4.7	62.1%	estimated ± 5.0 pp, medium confidence
24	Qwen3.8 Max Preview	62.1%	estimated ± 5.0 pp, medium confidence
25	Grok 4.20	62.0%	estimated ± 5.3 pp, low confidence
26	GPT-5.6 Terra	61.9%	estimated ± 5.0 pp, medium confidence
27	Grok 4.6	61.9%	estimated ± 5.0 pp, medium confidence
28	Gemini 3.5 Flash	61.7%	estimated ± 5.0 pp, medium confidence
29	Grok 4.5	61.7%	estimated ± 5.0 pp, medium confidence
30	GPT-5.3 Codex	61.6%	estimated ± 5.0 pp, medium confidence
31	Claude Opus 4.7 (Adaptive)	61.4%	estimated ± 5.0 pp, medium confidence
32	GLM-5.3	61.4%	estimated ± 5.0 pp, medium confidence
33	GPT-6 Luna	61.2%	measured
34	Claude Sonnet 5	60.5%	estimated ± 5.0 pp, medium confidence
35	GLM-5.2	60.4%	estimated ± 5.0 pp, medium confidence
36	DeepSeek V4 Pro 0813	60.3%	estimated ± 5.0 pp, medium confidence
37	Gemini 3.6 Flash	60.1%	estimated ± 5.0 pp, medium confidence
38	Muse Spark	60.0%	estimated ± 5.0 pp, medium confidence
39	Qwen3.7 Max	59.9%	estimated ± 5.0 pp, medium confidence
40	GPT-6 Sol	59.5%	measured
41	Claude Opus 4.6 (Adaptive)	59.3%	estimated ± 5.0 pp, medium confidence
42	GLM-5.3-Flash	59.3%	estimated ± 5.0 pp, medium confidence
43	Gemini 3 Pro	59.2%	estimated ± 5.0 pp, medium confidence
44	GPT-5.6 Luna	59.0%	estimated ± 5.0 pp, medium confidence
45	Ling 3.1 Flash	58.9%	estimated ± 5.0 pp, medium confidence
46	DeepSeek V4.1 Flash	58.7%	estimated ± 5.0 pp, medium confidence
47	MiniMax M3	58.6%	estimated ± 5.0 pp, medium confidence
48	DeepSeek V4 Flash 0731	58.2%	estimated ± 5.0 pp, medium confidence
49	Qwen3.8-Flash-Next	57.7%	estimated ± 5.0 pp, low confidence
50	GPT-5.2	57.4%	estimated ± 5.0 pp, low confidence
51	Kimi K2.6	57.2%	estimated ± 5.0 pp, low confidence
52	Grok 4.3	57.0%	estimated ± 5.0 pp, low confidence
53	GPT-5.2-Codex	55.6%	estimated ± 5.0 pp, low confidence
54	MiMo-V2.5-Pro	55.6%	estimated ± 5.0 pp, low confidence
55	Qwen3.7 Plus	55.5%	estimated ± 5.0 pp, low confidence
56	MiMo-V2.6-Flash	55.1%	estimated ± 5.0 pp, low confidence
57	Kimi K2.7 Code	55.0%	estimated ± 5.0 pp, low confidence
58	Mistral Large 4	55.0%	estimated ± 5.0 pp, low confidence
59	Apodex 1.1	54.2%	estimated ± 5.0 pp, low confidence
60	Apodex 1.1 Mini	54.2%	estimated ± 5.0 pp, low confidence
61	Qwen3.8-27B	54.0%	estimated ± 5.0 pp, low confidence
62	Hy3	53.6%	estimated ± 5.0 pp, low confidence
63	Hy3 Preview	53.6%	estimated ± 5.0 pp, low confidence
64	Claude Opus 4.7	53.4%	estimated ± 5.0 pp, low confidence
65	Inkling-Small	53.4%	estimated ± 5.0 pp, low confidence
66	Inkling	52.1%	estimated ± 5.0 pp, low confidence
67	Qwen 3.6 Max (preview)	51.1%	estimated ± 5.0 pp, low confidence
68	Kimi K2.5	51.0%	estimated ± 5.0 pp, low confidence
69	Kimi K2.5 (Reasoning)	51.0%	estimated ± 5.0 pp, low confidence
70	MiMo-V2-Pro	50.7%	estimated ± 5.0 pp, low confidence
71	Claude Opus 4.5 Thinking	50.4%	estimated ± 5.0 pp, low confidence
72	GLM-5.1	50.4%	estimated ± 5.0 pp, low confidence
73	A.X K2	49.9%	estimated ± 5.0 pp, low confidence
74	MiniMax M2.7	49.9%	estimated ± 5.0 pp, low confidence
75	GLM-5	49.6%	estimated ± 5.0 pp, low confidence
76	Solar Pro 4	49.5%	estimated ± 5.0 pp, low confidence
77	GPT-5.1	48.8%	estimated ± 5.0 pp, low confidence
78	GPT-5 (high)	48.8%	estimated ± 5.0 pp, low confidence
79	Nemotron 3 Ultra	48.7%	estimated ± 5.0 pp, low confidence
80	GPT-5.4 nano	48.6%	estimated ± 5.0 pp, low confidence
81	GPT-5.4 mini	48.4%	estimated ± 5.0 pp, low confidence
82	GLM-5-Turbo	48.1%	estimated ± 5.0 pp, low confidence
83	Qwen3.6 Plus	48.1%	estimated ± 5.0 pp, low confidence
84	GLM-4.7	47.7%	estimated ± 5.0 pp, low confidence
85	Grok 4	47.0%	estimated ± 5.0 pp, low confidence
86	GPT-5.1-Codex	46.0%	estimated ± 5.0 pp, low confidence
87	GPT-5.1-Codex-Max	46.0%	estimated ± 5.0 pp, low confidence
88	GPT-5 (medium)	45.7%	estimated ± 5.0 pp, low confidence
89	Qwen3.5-122B-A10B	45.5%	estimated ± 5.0 pp, low confidence
90	Qwen3.5-27B	44.2%	estimated ± 5.0 pp, low confidence
91	Ling 3.0 Flash	44.0%	estimated ± 5.0 pp, low confidence
92	Ling 3.0 Flash FP8	44.0%	estimated ± 5.0 pp, low confidence
93	Gemma 4 31B	43.9%	estimated ± 5.0 pp, low confidence
94	Qwen3.6-27B	43.3%	estimated ± 5.0 pp, low confidence
95	Gemini 2.5 Pro	42.7%	estimated ± 5.0 pp, low confidence
96	Qwen3.6-35B-A3B	42.4%	estimated ± 5.0 pp, low confidence
97	MiMo-V2-Omni	42.3%	estimated ± 5.0 pp, low confidence
98	Ling 3.0 Flash VL	42.2%	estimated ± 5.0 pp, low confidence
99	Muse Glimmer 30B	42.2%	estimated ± 5.0 pp, low confidence
100	Step 3.7 Flash	41.5%	estimated ± 5.0 pp, low confidence
101	Qwen3.5-35B-A3B	41.1%	estimated ± 5.0 pp, low confidence
102	Nemotron 3 Super 100B	40.9%	estimated ± 5.0 pp, low confidence
103	o3-pro	40.8%	estimated ± 5.0 pp, low confidence
104	o3	40.1%	estimated ± 5.0 pp, low confidence
105	Qwen3.5 397B	39.8%	estimated ± 5.0 pp, low confidence
106	Qwen3.5 397B (Reasoning)	39.8%	estimated ± 5.0 pp, low confidence
107	GPT-OSS 120B	39.6%	estimated ± 5.0 pp, low confidence
108	Gemma 4 26B A4B	39.3%	estimated ± 5.0 pp, low confidence
109	Grok 4.1 Fast (Reasoning)	39.3%	estimated ± 5.0 pp, low confidence
110	Claude Opus 4.6	39.0%	estimated ± 5.0 pp, low confidence
111	Grok 4 Fast (Reasoning)	39.0%	estimated ± 5.0 pp, low confidence
112	Gemini 3.5 Flash-Lite	38.7%	estimated ± 5.0 pp, low confidence
113	Quasar 438B	38.6%	estimated ± 5.0 pp, low confidence
114	K-EXAONE 2.0	38.5%	estimated ± 5.0 pp, low confidence
115	Claude 4.1 Opus	36.8%	estimated ± 5.0 pp, low confidence
116	GLM-5V-Turbo	36.8%	estimated ± 5.0 pp, low confidence
117	DeepSeek-R1	35.3%	estimated ± 5.0 pp, low confidence
118	Trinity-Large-Preview	35.3%	estimated ± 5.0 pp, low confidence
119	Trinity-Large-Thinking	35.3%	estimated ± 5.0 pp, low confidence
120	Gemma 4 12B	35.2%	estimated ± 5.0 pp, low confidence
121	Gemini 3 Flash	34.4%	estimated ± 5.0 pp, low confidence
122	DeepSeek V3.1 (Reasoning)	33.6%	estimated ± 5.0 pp, low confidence
123	K-Exaone	33.1%	estimated ± 5.0 pp, low confidence
124	Mistral Medium 3.5 128B	33.0%	estimated ± 5.0 pp, low confidence
125	Claude Sonnet 4.6	32.4%	estimated ± 5.0 pp, low confidence
126	Claude Opus 4.5	32.3%	estimated ± 5.0 pp, low confidence
127	Claude 4.1 Opus Thinking	31.4%	estimated ± 5.0 pp, low confidence
128	Command A+	30.8%	estimated ± 5.0 pp, low confidence
129	Qwen3 Max	30.7%	estimated ± 5.0 pp, low confidence
130	Mercury 2.5	30.6%	estimated ± 5.0 pp, low confidence
131	Nemotron 3 Nano 30B	30.1%	estimated ± 5.0 pp, low confidence
132	DeepSeek V3.2	29.9%	estimated ± 5.0 pp, low confidence
133	Granite 4.2 30B	29.9%	estimated ± 5.0 pp, low confidence
134	North Mini Code	29.7%	estimated ± 5.0 pp, low confidence
135	GPT-OSS 20B	29.6%	estimated ± 5.0 pp, low confidence
136	Sarvam 105B	29.6%	estimated ± 5.0 pp, low confidence
137	Nemotron 3.5 Lightning 30B A3B NVFP4	29.1%	estimated ± 5.0 pp, low confidence
138	o1-pro	28.8%	estimated ± 5.0 pp, low confidence
139	Solar Pro 3	28.8%	estimated ± 5.0 pp, low confidence
140	Mistral Small 4	28.3%	estimated ± 5.0 pp, low confidence
141	Mistral Small 4 (Reasoning)	28.3%	estimated ± 5.0 pp, low confidence
142	Granite 4.2 8B	28.0%	estimated ± 5.0 pp, low confidence
143	Ling 3.0 Tiny	27.5%	estimated ± 5.0 pp, low confidence
144	o1-preview	27.4%	estimated ± 5.0 pp, low confidence
145	MiniCPM5-2B	27.0%	estimated ± 5.0 pp, low confidence
146	MiMo-V2-Flash	26.6%	estimated ± 5.0 pp, low confidence
147	Grok Code Fast 1	25.9%	estimated ± 5.0 pp, low confidence
148	o3-mini	25.7%	estimated ± 5.0 pp, low confidence
149	Qwen3-Omni-30B-A3B-Thinking	25.2%	estimated ± 5.0 pp, low confidence
150	Sarvam 30B	25.2%	estimated ± 5.0 pp, low confidence
151	Kimi K2	25.1%	estimated ± 5.0 pp, low confidence
152	Nemotron Ultra 253B	25.1%	estimated ± 5.0 pp, low confidence
153	GLM-4.5-Air	24.6%	estimated ± 5.0 pp, low confidence
154	o1	24.6%	estimated ± 5.0 pp, low confidence
155	LFM2.5-8B-A1B	24.5%	estimated ± 5.0 pp, low confidence
156	Celeris-1	24.3%	estimated ± 5.0 pp, low confidence
157	DeepSeek V3.1	24.2%	estimated ± 5.0 pp, low confidence
158	Granite 4.2 3B	24.1%	estimated ± 5.0 pp, low confidence
159	Granite-4.0-H-350M	23.8%	estimated ± 5.0 pp, low confidence
160	Ling 2.6 Flash	23.7%	estimated ± 5.0 pp, low confidence
161	LFM2.5-2.6B	23.6%	estimated ± 5.0 pp, low confidence
162	Exaone 4.0 1.2B	22.9%	estimated ± 5.0 pp, low confidence
163	GLM-4.6	22.6%	estimated ± 5.0 pp, low confidence
164	Granite-4.0-350M	22.6%	estimated ± 5.0 pp, low confidence
165	Grok 4.1 Fast	22.1%	estimated ± 5.0 pp, low confidence
166	LFM2.5-VL-1.6B-Extract	22.1%	estimated ± 5.0 pp, low confidence
167	Exaone 4.0 32B	22.0%	estimated ± 5.0 pp, low confidence
168	GPT-4.1 mini	22.0%	estimated ± 5.0 pp, low confidence
169	Granite-4.0-H-1B	22.0%	estimated ± 5.0 pp, low confidence
170	Phi-4 Multimodal Instruct	22.0%	estimated ± 5.0 pp, low confidence
171	Llama 4 Maverick	21.8%	estimated ± 5.0 pp, low confidence
172	Gemma 4 E2B	21.7%	estimated ± 5.0 pp, low confidence
173	Nemotron 3 Nano Omni 30B A3B	21.7%	estimated ± 5.0 pp, low confidence
174	DeepSeek V3 0324	21.6%	estimated ± 5.0 pp, low confidence
175	Gemini 2.5 Flash	21.6%	estimated ± 5.0 pp, low confidence
176	DeepSeek R1 Distill Qwen 32B	21.4%	estimated ± 5.0 pp, low confidence
177	Gemini 1.5 Pro	21.4%	estimated ± 5.0 pp, low confidence
178	Qwen3-Omni-30B-A3B-Instruct	21.4%	estimated ± 5.0 pp, low confidence
179	Gemma 3 27B	21.2%	estimated ± 5.0 pp, low confidence
180	Claude 4 Sonnet	21.0%	estimated ± 5.0 pp, low confidence
181	Gemini 1.0 Pro	20.9%	estimated ± 5.0 pp, low confidence
182	GPT-4.1	20.9%	estimated ± 5.0 pp, low confidence
183	GPT-4o mini	20.9%	estimated ± 5.0 pp, low confidence
184	Mistral Large 3	20.9%	estimated ± 5.0 pp, low confidence
185	Claude 3 Haiku	20.8%	estimated ± 5.0 pp, low confidence
186	Mistral Medium 3	20.8%	estimated ± 5.0 pp, low confidence
187	Llama 3.1 405B	20.6%	estimated ± 5.0 pp, low confidence
188	Gemma 4 E4B	20.4%	estimated ± 5.0 pp, low confidence
189	GPT-4.1 nano	20.4%	estimated ± 5.0 pp, low confidence
190	Llama 4 Scout	20.4%	estimated ± 5.0 pp, low confidence
191	Phi-4	20.4%	estimated ± 5.0 pp, low confidence
192	Solar Pro 2	20.2%	estimated ± 5.0 pp, low confidence
193	Ultravox v0.6 Llama 3.3 70B	20.1%	estimated ± 5.0 pp, low confidence
194	Qwen2.5 Coder 32B Instruct	20.0%	estimated ± 5.0 pp, low confidence
195	Mistral Large 2	19.7%	estimated ± 5.0 pp, low confidence
196	Nova Pro	19.6%	estimated ± 5.0 pp, low confidence
197	GPT-4 Turbo	19.4%	estimated ± 5.0 pp, low confidence
198	DeepSeek V3	19.1%	estimated ± 5.0 pp, low confidence
199	Claude 3 Opus	19.0%	estimated ± 5.0 pp, low confidence
200	GPT-4o	18.5%	estimated ± 5.0 pp, low confidence

HealthBench Professional (raw) leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General