Math

HMMT Nov 2025 leaderboard

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

As of 2026-10-07, the highest measured score on HMMT Nov 2025 is 96.9% by GLM-5. 43 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Claude 3.5 Sonnet	99.9%	estimated ± 2.6 pp, low confidence
2	GLM-4.7	99.9%	estimated ± 2.6 pp, low confidence
3	GPT-4.1	99.9%	estimated ± 2.6 pp, low confidence
4	Grok 3 [Beta]	99.9%	estimated ± 2.6 pp, low confidence
5	Kimi K2	99.9%	estimated ± 2.6 pp, low confidence
6	Qwen3 235B 2507 (Reasoning)	99.9%	estimated ± 2.6 pp, low confidence
7	Qwen3.5 Flash	99.9%	estimated ± 2.6 pp, low confidence
8	GLM-5	96.9%	measured
9	GPT-5.4 mini	96.8%	estimated ± 2.6 pp, low confidence
10	Claude Haiku 4.5	96.8%	estimated ± 2.6 pp, low confidence
11	Grok 4	96.8%	estimated ± 2.6 pp, low confidence
12	o3	96.8%	estimated ± 2.6 pp, low confidence
13	Qwen3.5 Plus	96.8%	estimated ± 2.6 pp, low confidence
14	DeepSeek V3.2	96.8%	estimated ± 2.6 pp, medium confidence
15	GLM-4.6	96.7%	estimated ± 2.6 pp, medium confidence
16	Qwen3.6 Plus	94.6%	measured
17	Kimi K2.6	94.4%	estimated ± 2.0 pp, high confidence
18	GLM-5.2	94.4%	measured
19	GLM-5.1	94.0%	measured
20	Claude Sonnet 4.5	93.3%	estimated ± 2.6 pp, medium confidence
21	Gemini 2.5 Flash	93.3%	estimated ± 2.6 pp, medium confidence
22	Gemini 2.5 Pro	93.3%	estimated ± 2.6 pp, medium confidence
23	Gemini 3 Flash	93.3%	estimated ± 2.6 pp, medium confidence
24	Qwen 3.6 Max (preview)	93.3%	estimated ± 2.6 pp, medium confidence
25	Claude Opus 4.5	93.3%	measured
26	GPT-5.4 nano	93.2%	estimated ± 2.6 pp, medium confidence
27	o4-mini (high)	93.2%	estimated ± 2.6 pp, medium confidence
28	Claude Opus 4.6	93.2%	estimated ± 2.6 pp, low confidence
29	Claude Opus 4.7	93.2%	estimated ± 2.6 pp, low confidence
30	Claude Opus 4.8	93.2%	estimated ± 2.6 pp, low confidence
31	Claude Sonnet 4.6	93.2%	estimated ± 2.6 pp, medium confidence
32	Gemini 3.1 Pro	93.2%	estimated ± 2.6 pp, low confidence
33	Gemini 3.5 Flash	93.2%	estimated ± 2.6 pp, low confidence
34	Gemini 3 Pro	93.2%	estimated ± 2.6 pp, low confidence
35	GPT-5.1	93.2%	estimated ± 2.6 pp, medium confidence
36	GPT-5.2	93.2%	estimated ± 2.6 pp, low confidence
37	GPT-5.4	93.2%	estimated ± 2.6 pp, low confidence
38	GPT-5.4 Pro	93.2%	estimated ± 2.6 pp, low confidence
39	GPT-5.5	93.2%	estimated ± 2.6 pp, low confidence
40	GPT-5.5 Pro	93.2%	estimated ± 2.6 pp, low confidence
41	GPT-5.6 Luna	93.2%	estimated ± 2.6 pp, low confidence
42	GPT-5.6 Sol	93.2%	estimated ± 2.6 pp, low confidence
43	GPT-5.6 Terra	93.2%	estimated ± 2.6 pp, low confidence
44	GPT-6 Astra	93.2%	estimated ± 2.6 pp, low confidence
45	Muse Spark	93.2%	estimated ± 2.6 pp, low confidence
46	Qwen3.5 397B	92.7%	measured
47	Kimi K2.5	91.1%	measured
48	Qwen3.6-27B	90.7%	measured
49	Qwen3.6-35B-A3B	89.1%	measured
50	Granite 4.2 30B	87.7%	estimated ± 1.9 pp, low confidence
51	Granite 4.2 8B	77.8%	estimated ± 1.9 pp, low confidence
52	Granite 4.2 3B	67.1%	estimated ± 1.9 pp, low confidence

HMMT Nov 2025 leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General