Vision & documents

MedXpertQA (MM) leaderboard

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

As of 2026-10-07, the highest measured score on MedXpertQA (MM) is 81.3% by Gemini 3.1 Pro. 114 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Gemini 3.5 Flash	81.7%	estimated ± 2.4 pp, high confidence
2	Gemini 3.1 Pro	81.3%	measured
3	GPT-5.6 Sol	80.6%	estimated ± 2.4 pp, high confidence
4	Claude Mythos 5	80.4%	estimated ± 6.4 pp, low confidence
5	Qwen3.8 Max	80.4%	measured
6	Claude Opus 4.7 (Adaptive)	79.4%	estimated ± 6.4 pp, low confidence
7	Claude Opus 4.8	79.0%	estimated ± 6.4 pp, low confidence
8	GLM-5.3-Flash	78.8%	estimated ± 6.4 pp, low confidence
9	Qwen3.8-Flash-Next	78.5%	estimated ± 4.2 pp, medium confidence
10	Gemini 3.7 Flash	78.5%	estimated ± 6.4 pp, low confidence
11	Muse Spark	78.4%	measured
12	Muse Spark 1.1	78.4%	estimated ± 6.4 pp, low confidence
13	Claude Sonnet 5	78.4%	estimated ± 6.4 pp, low confidence
14	Qwen3.8-Omni-Flash	78.2%	estimated ± 4.2 pp, medium confidence
15	Kimi K3	77.9%	estimated ± 2.4 pp, high confidence
16	Seed 2.1 Pro	77.9%	estimated ± 2.4 pp, high confidence
17	Claude Opus 5.5	77.9%	estimated ± 6.5 pp, low confidence
18	GPT-6 Astra	77.8%	estimated ± 6.5 pp, low confidence
19	GPT-6.1 Sol	77.8%	estimated ± 6.5 pp, low confidence
20	Gemini 3.8 Flash	77.8%	estimated ± 6.5 pp, low confidence
21	Claude Opus 5	77.8%	estimated ± 6.5 pp, low confidence
22	Sakana Fugu-Ultra	77.7%	estimated ± 6.4 pp, low confidence
23	Gemini 3.6 Flash	77.6%	estimated ± 6.5 pp, low confidence
24	GPT-6 Sol	77.6%	estimated ± 6.5 pp, low confidence
25	Qwen3.8 Max Preview	77.6%	estimated ± 6.5 pp, low confidence
26	GPT-5.4	77.1%	measured
27	Grok 4.5	77.1%	estimated ± 6.5 pp, low confidence
28	Sakana Fugu	77.1%	estimated ± 6.4 pp, low confidence
29	GPT-5.5	77.1%	estimated ± 2.4 pp, high confidence
30	GPT-6 Luna	76.9%	estimated ± 6.5 pp, low confidence
31	Gemini 3 Pro	76.6%	estimated ± 2.4 pp, high confidence
32	Apodex 1.1	76.6%	estimated ± 6.5 pp, low confidence
33	Apodex 1.1 Mini	76.6%	estimated ± 6.5 pp, low confidence
34	Gemini 3.5 Flash-Lite	76.5%	estimated ± 6.5 pp, low confidence
35	Ling 3.0 Flash VL	76.5%	estimated ± 6.5 pp, low confidence
36	Gemini 3 Flash	76.3%	estimated ± 6.5 pp, low confidence
37	Qwen3.8-27B	76.3%	estimated ± 4.2 pp, medium confidence
38	GPT-5.3 Codex	76.3%	estimated ± 6.5 pp, low confidence
39	GPT-5.6 Terra	76.0%	estimated ± 2.4 pp, high confidence
40	DeepSeek V4.1 Flash	75.0%	estimated ± 6.5 pp, low confidence
41	Seed 2.1 Turbo	74.8%	estimated ± 2.4 pp, high confidence
42	Claude Opus 4.7	74.3%	estimated ± 6.5 pp, low confidence
43	Mistral Large 4	74.3%	estimated ± 6.5 pp, low confidence
44	GPT-5.2-Codex	74.2%	estimated ± 6.5 pp, low confidence
45	Claude Sonnet 4.6	74.0%	estimated ± 6.4 pp, low confidence
46	Qwen3.5-122B-A10B	73.9%	estimated ± 6.4 pp, low confidence
47	Nemotron 3 Nano Omni 30B A3B	73.5%	estimated ± 6.4 pp, low confidence
48	GPT-5.2	73.5%	estimated ± 2.4 pp, high confidence
49	Kimi K2.6	73.3%	estimated ± 2.4 pp, high confidence
50	GPT-5.1	73.0%	estimated ± 6.5 pp, low confidence
51	Claude Opus 4.6 (Adaptive)	72.8%	estimated ± 6.5 pp, low confidence
52	Step 3.7 Flash	72.6%	estimated ± 6.5 pp, low confidence
53	dots3-note Preview	72.6%	estimated ± 2.4 pp, high confidence
54	Qwen3.5 397B	72.4%	estimated ± 2.4 pp, high confidence
55	Gemini 3.1 Flash-Lite	72.3%	estimated ± 6.4 pp, low confidence
56	Qwen3.5-27B	72.1%	estimated ± 6.5 pp, low confidence
57	Qwen3.6 Plus	72.0%	estimated ± 2.4 pp, high confidence
58	Gemini 2.5 Pro	71.9%	estimated ± 6.5 pp, low confidence
59	Kimi K2.5	71.3%	estimated ± 2.4 pp, high confidence
60	Kimi K2.5 (Reasoning)	71.3%	estimated ± 2.4 pp, high confidence
61	GPT-5.6 Luna	71.1%	estimated ± 2.4 pp, high confidence
62	Qwen3.7 Plus	71.0%	measured
63	GPT-5 (medium)	70.5%	estimated ± 6.5 pp, low confidence
64	Grok 4.3	70.4%	estimated ± 2.4 pp, high confidence
65	MiniMax M3	70.4%	estimated ± 2.4 pp, high confidence
66	GPT-5 (high)	70.2%	estimated ± 6.5 pp, low confidence
67	Pareto 26.9	70.2%	estimated ± 2.4 pp, high confidence
68	MiMo-V2.5	70.0%	estimated ± 2.4 pp, high confidence
69	Claude Opus 4.5 Thinking	69.7%	estimated ± 6.5 pp, low confidence
70	Gemma 4 31B	67.7%	estimated ± 2.4 pp, high confidence
71	GPT-5.4 mini	67.0%	estimated ± 2.4 pp, high confidence
72	MiMo-V2.6-Flash	66.8%	estimated ± 6.5 pp, low confidence
73	Grok 4.20	65.8%	measured
74	GLM-5V-Turbo	65.7%	estimated ± 6.5 pp, low confidence
75	Step 5 Preview	65.6%	estimated ± 2.4 pp, high confidence
76	Qwen3.5-35B-A3B	65.3%	estimated ± 6.5 pp, low confidence
77	Qwen3.6-27B	65.1%	estimated ± 2.4 pp, high confidence
78	Claude Opus 4.6	64.8%	measured
79	GPT-5.1-Codex	64.5%	estimated ± 6.5 pp, low confidence
80	GPT-5.1-Codex-Max	64.5%	estimated ± 6.5 pp, low confidence
81	Qwen3.6-35B-A3B	64.0%	estimated ± 2.4 pp, high confidence
82	Inkling-Small	60.8%	estimated ± 2.4 pp, high confidence
83	Muse Glimmer 30B	60.8%	estimated ± 2.4 pp, high confidence
84	Gemma 4 26B A4B	60.3%	estimated ± 2.4 pp, high confidence
85	Inkling	59.6%	estimated ± 2.4 pp, high confidence
86	Interfaze Beta	53.7%	estimated ± 2.4 pp, high confidence
87	Claude Opus 4.5	52.4%	estimated ± 2.4 pp, high confidence
88	o3	51.5%	estimated ± 6.5 pp, low confidence
89	MiMo-V2-Omni	50.2%	estimated ± 6.5 pp, low confidence
90	Gemma 4 12B	48.7%	measured
91	Grok 4	42.5%	estimated ± 6.5 pp, low confidence
92	GPT-5.4 nano	41.2%	estimated ± 2.4 pp, medium confidence
93	Claude 4.1 Opus Thinking	35.9%	estimated ± 6.5 pp, low confidence
94	Command A+	33.8%	estimated ± 2.4 pp, medium confidence
95	Gemini 2.5 Flash	20.1%	estimated ± 6.5 pp, low confidence
96	Mistral Medium 3.5 128B	16.9%	estimated ± 6.5 pp, low confidence
97	Grok 4.1 Fast (Reasoning)	10.3%	estimated ± 6.5 pp, low confidence
98	Claude 4 Sonnet	7.6%	estimated ± 6.5 pp, low confidence
99	Llama 4 Maverick	6.9%	estimated ± 6.5 pp, low confidence
100	Grok 4 Fast (Reasoning)	6.2%	estimated ± 6.5 pp, low confidence
101	GPT-4.1	5.0%	estimated ± 6.5 pp, low confidence
102	Qwen3-Omni-30B-A3B-Thinking	3.5%	estimated ± 6.5 pp, low confidence
103	GPT-4.1 mini	2.0%	estimated ± 6.5 pp, low confidence
104	Mistral Small 4	1.0%	estimated ± 6.5 pp, low confidence
105	Mistral Small 4 (Reasoning)	1.0%	estimated ± 6.5 pp, low confidence
106	LFM2.5-VL-3B	0.7%	estimated ± 2.4 pp, medium confidence
107	Mistral Large 3	0.7%	estimated ± 6.5 pp, low confidence
108	Qwen3-Omni-30B-A3B-Instruct	0.6%	estimated ± 6.5 pp, low confidence
109	Gemini 1.5 Pro	0.5%	estimated ± 6.5 pp, low confidence
110	Mistral Medium 3	0.2%	estimated ± 6.5 pp, low confidence
111	Llama 4 Scout	0.2%	estimated ± 6.5 pp, low confidence
112	Qwen3.5 397B (Reasoning)	0.2%	estimated ± 6.5 pp, low confidence
113	Gemma 4 E4B	0.1%	estimated ± 6.5 pp, low confidence
114	Grok 4.1 Fast	0.0%	estimated ± 6.5 pp, low confidence
115	Gemma 3 27B	0.0%	estimated ± 6.5 pp, low confidence
116	Gemma 4 E2B	0.0%	estimated ± 6.5 pp, low confidence
117	Nova Pro	0.0%	estimated ± 6.5 pp, low confidence
118	GPT-4o mini	0.0%	estimated ± 6.5 pp, low confidence
119	GPT-4.1 nano	0.0%	estimated ± 6.5 pp, low confidence
120	Claude 3 Haiku	0.0%	estimated ± 6.5 pp, low confidence
121	LFM2.5-VL-1.6B-Extract	0.0%	estimated ± 6.5 pp, low confidence
122	Phi-4 Multimodal Instruct	0.0%	estimated ± 6.5 pp, low confidence

MedXpertQA (MM) leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General