Math

HMMT Feb 2025 leaderboard

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

As of 2026-10-07, the highest measured score on HMMT Feb 2025 is 97.5% by GLM-5. 26 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Qwen3.7 Max	100.0%	estimated ± 1.8 pp, low confidence
2	DeepSeek V4 Pro 0813	99.6%	estimated ± 1.8 pp, low confidence
3	DeepSeek V4 Flash 0731	99.4%	estimated ± 1.8 pp, low confidence
4	Solar Open 2	98.9%	estimated ± 1.8 pp, low confidence
5	Qwen3.7 Plus	98.4%	estimated ± 1.8 pp, low confidence
6	Beam	98.4%	estimated ± 1.9 pp, low confidence
7	Kimi K2.6	98.3%	estimated ± 1.8 pp, low confidence
8	A.X K2	97.6%	estimated ± 1.9 pp, low confidence
9	Inkling	97.6%	estimated ± 1.9 pp, low confidence
10	GLM-5	97.5%	measured
11	Inkling-Small	96.9%	estimated ± 1.8 pp, low confidence
12	Qwen3.6 Plus	96.7%	measured
13	Ternary Bonsai 2 27B	96.1%	estimated ± 1.9 pp, low confidence
14	GLM-5.2	95.5%	estimated ± 1.6 pp, medium confidence
15	Solar Pro 4	95.5%	estimated ± 1.9 pp, medium confidence
16	Kimi K2.5	95.4%	measured
17	GLM-5.1	95.3%	estimated ± 1.6 pp, medium confidence
18	Ling 3.0 Flash	95.1%	estimated ± 1.8 pp, low confidence
19	Qwen3.5 397B	94.8%	measured
20	Muse Glimmer 30B	94.7%	estimated ± 1.9 pp, medium confidence
21	MAI-Thinking-1	93.9%	estimated ± 1.8 pp, low confidence
22	Qwen3.6-27B	93.8%	measured
23	Claude Opus 4.5	92.9%	measured
24	Qwen3.6-35B-A3B	90.7%	measured
25	K-EXAONE 2.0	89.9%	estimated ± 1.8 pp, low confidence
26	Granite 4.2 30B	89.2%	measured
27	ZAYA1-8B	85.5%	estimated ± 1.8 pp, low confidence
28	MiniCPM5-2B	79.9%	estimated ± 1.8 pp, low confidence
29	Granite 4.2 8B	78.3%	measured
30	Granite 4.2 3B	66.7%	measured
31	Gemma 4 12B	63.5%	estimated ± 1.9 pp, low confidence
32	ZAYA1-74B-Preview	61.0%	estimated ± 1.9 pp, low confidence
33	LongCat-Flash-Lite-Sparse	59.4%	estimated ± 1.8 pp, low confidence
34	MiniCPM5-1B	42.3%	estimated ± 1.8 pp, low confidence
35	LFM2.5-8B-A1B	9.0%	estimated ± 1.9 pp, low confidence
36	LLaDA2.2-mini	1.1%	estimated ± 1.9 pp, low confidence

HMMT Feb 2025 leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General