Knowledge & reasoning

MedXpertQA (Text) leaderboard

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

As of 2026-10-07, the highest measured score on MedXpertQA (Text) is 71.5% by Gemini 3.1 Pro. 78 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	78.7%	estimated ± 8.9 pp, low confidence
2	GPT-6 Astra	77.7%	estimated ± 4.5 pp, low confidence
3	Claude Fable 5.1	76.1%	estimated ± 8.9 pp, low confidence
4	Sakana Fugu	75.1%	estimated ± 4.5 pp, low confidence
5	Sakana Fugu-Ultra	75.1%	estimated ± 4.5 pp, low confidence
6	Claude Mythos 5	74.6%	estimated ± 8.9 pp, low confidence
7	Claude Sonnet 5.5	72.9%	estimated ± 8.9 pp, low confidence
8	Claude Opus 5	72.4%	estimated ± 8.9 pp, low confidence
9	Gemini 3.1 Pro	71.5%	measured
10	GPT-5.6 Sol	70.7%	estimated ± 4.5 pp, low confidence
11	Muse Spark 1.1	68.9%	estimated ± 8.9 pp, low confidence
12	Claude Opus 4.7 (Adaptive)	68.9%	estimated ± 4.5 pp, medium confidence
13	Claude Opus 4.8	66.3%	estimated ± 4.5 pp, medium confidence
14	GPT-5.5	66.3%	estimated ± 4.5 pp, medium confidence
15	Pareto 26.9	66.1%	estimated ± 8.9 pp, low confidence
16	Kimi K3	65.9%	estimated ± 4.5 pp, medium confidence
17	Step 5 Preview	65.9%	estimated ± 4.5 pp, medium confidence
18	GPT-5.6 Terra	63.5%	estimated ± 4.5 pp, medium confidence
19	Claude Haiku 5.5	63.2%	estimated ± 8.9 pp, low confidence
20	Ornith-1.5-397B	63.2%	estimated ± 4.5 pp, medium confidence
21	Gemini 3.5 Flash	62.7%	estimated ± 4.5 pp, medium confidence
22	Qwen3.8 Max	62.4%	estimated ± 4.5 pp, medium confidence
23	Pareto 26.10 Preview	61.7%	estimated ± 4.5 pp, medium confidence
24	Qwen3.7 Max	61.7%	estimated ± 4.5 pp, medium confidence
25	GPT-5.6 Luna	61.3%	estimated ± 4.5 pp, medium confidence
26	Hy4 preview	61.3%	estimated ± 4.5 pp, medium confidence
27	Claude Sonnet 5	60.7%	estimated ± 8.9 pp, low confidence
28	GPT-5.5 Pro	60.6%	estimated ± 8.9 pp, low confidence
29	GPT-5.4 Pro	60.2%	estimated ± 8.9 pp, low confidence
30	GPT-5.4	59.6%	measured
31	Qwen3.8-Flash-Next	59.2%	estimated ± 4.5 pp, medium confidence
32	GLM-5.2	57.5%	estimated ± 4.5 pp, medium confidence
33	Qwen3.8-Omni-Flash	56.8%	estimated ± 4.5 pp, medium confidence
34	DeepSeek V4.1 Flash	56.5%	estimated ± 4.5 pp, medium confidence
35	Kimi K2.6	55.3%	estimated ± 4.5 pp, medium confidence
36	Beam	55.3%	estimated ± 4.5 pp, medium confidence
37	Qwen3.7 Plus	54.7%	estimated ± 4.5 pp, medium confidence
38	DeepSeek V4 Pro 0813	54.1%	estimated ± 4.5 pp, medium confidence
39	Interfaze Beta	53.5%	estimated ± 4.5 pp, medium confidence
40	Muse Spark	52.6%	measured
41	Inkling-Small	52.3%	estimated ± 4.5 pp, medium confidence
42	Claude Opus 4.6	52.1%	measured
43	Ornith-1.5-35B-A3B	51.5%	estimated ± 4.5 pp, medium confidence
44	Qwen3.8-27B	51.5%	estimated ± 4.5 pp, medium confidence
45	MiMo-V2.5-Pro	51.0%	estimated ± 8.9 pp, low confidence
46	Solar Pro 4	50.9%	estimated ± 4.5 pp, medium confidence
47	Grok 4.20	50.2%	measured
48	DeepSeek V4 Flash 0731	48.6%	estimated ± 4.5 pp, low confidence
49	Inkling	48.1%	estimated ± 4.5 pp, low confidence
50	Kimi K2.5	47.4%	estimated ± 4.5 pp, low confidence
51	Hy3 Preview	46.4%	estimated ± 4.5 pp, low confidence
52	MiniMax M2.7	45.9%	estimated ± 4.5 pp, low confidence
53	Nemotron 3 Ultra	45.9%	estimated ± 4.5 pp, low confidence
54	Ornith-1.5-9B	44.6%	estimated ± 4.5 pp, low confidence
55	Solar Open 2	44.4%	estimated ± 4.5 pp, low confidence
56	GPT-5.4 mini	44.3%	estimated ± 8.9 pp, low confidence
57	GLM-5.1	44.2%	estimated ± 4.5 pp, low confidence
58	GLM-5	43.7%	estimated ± 4.5 pp, low confidence
59	Ternary Bonsai 2 27B	43.2%	estimated ± 4.5 pp, low confidence
60	A.X K2	42.9%	estimated ± 4.5 pp, low confidence
61	Ling 3.0 Flash	41.6%	estimated ± 4.5 pp, low confidence
62	MAI-Thinking-1	40.1%	estimated ± 4.5 pp, low confidence
63	Ling 3.0 Flash FP8	39.8%	estimated ± 4.5 pp, low confidence
64	GPT-5.4 nano	39.3%	estimated ± 8.9 pp, low confidence
65	K-EXAONE 2.0	36.6%	estimated ± 4.5 pp, low confidence
66	Gemma 4 31B	32.8%	estimated ± 8.9 pp, low confidence
67	Mercury 2.5	31.9%	estimated ± 4.5 pp, low confidence
68	Gemma 4 12B	31.6%	estimated ± 4.5 pp, low confidence
69	Trinity-Large-Thinking	28.5%	estimated ± 4.5 pp, low confidence
70	Nemotron 3.5 Lightning 30B A3B NVFP4	27.7%	estimated ± 4.5 pp, low confidence
71	Nemotron 3 Nano Omni 30B A3B	24.3%	estimated ± 4.5 pp, low confidence
72	ZAYA1-8B	23.2%	estimated ± 4.5 pp, low confidence
73	MiniCPM5-2B	22.5%	estimated ± 4.5 pp, low confidence
74	LongCat-Flash-Lite-Sparse	21.9%	estimated ± 4.5 pp, low confidence
75	Trinity-Large-Preview	17.5%	estimated ± 4.5 pp, low confidence
76	Gemma 4 26B A4B	16.1%	estimated ± 8.9 pp, low confidence
77	Mellum2-12B-A2.5B-Thinking	14.2%	estimated ± 4.5 pp, low confidence
78	ZAYA1-74B-Preview	14.1%	estimated ± 4.5 pp, low confidence
79	LLaDA2.2-mini	8.9%	estimated ± 4.5 pp, low confidence
80	Soofi S 30B-A3B	8.5%	estimated ± 4.5 pp, low confidence
81	Mellum2-12B-A2.5B-Instruct	7.8%	estimated ± 4.5 pp, low confidence
82	MiniCPM5-1B	4.1%	estimated ± 4.5 pp, low confidence
83	LFM2.5-230M	4.0%	estimated ± 4.5 pp, low confidence

MedXpertQA (Text) leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General