Vision & documents

MathVision w/ Python leaderboard

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

As of 2026-10-07, the highest measured score on MathVision w/ Python is 97.8% by Kimi K3. 39 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Kimi K3	97.8%	measured
2	Qwen3.8 Max	97.7%	measured
3	Claude Mythos 5	97.7%	estimated ± 0.9 pp, medium confidence
4	Gemini 3.8 Flash	97.0%	estimated ± 1.3 pp, low confidence
5	Claude Opus 4.7 (Adaptive)	96.6%	estimated ± 0.9 pp, medium confidence
6	Seed 2.1 Pro	96.5%	estimated ± 0.5 pp, medium confidence
7	Qwen3.8-Omni-Flash	96.2%	measured
8	Qwen3.8-Flash-Next	95.7%	measured
9	Qwen3.7 Plus	95.3%	estimated ± 0.5 pp, medium confidence
10	Seed 2.1 Turbo	95.2%	estimated ± 0.5 pp, medium confidence
11	Qwen3.8-27B	94.6%	measured
12	Qwen3.5 397B	94.3%	estimated ± 0.5 pp, low confidence
13	Qwen3.6 Plus	94.0%	estimated ± 0.5 pp, low confidence
14	dots3-note Preview	93.8%	estimated ± 0.5 pp, low confidence
15	Kimi K2.6	93.6%	estimated ± 0.5 pp, low confidence
16	Claude Opus 4.8	93.3%	estimated ± 0.9 pp, low confidence
17	Gemini 3 Pro	93.2%	estimated ± 0.5 pp, low confidence
18	Qwen3.5-122B-A10B	93.0%	estimated ± 0.5 pp, low confidence
19	Qwen3.5-27B	92.8%	estimated ± 0.5 pp, low confidence
20	Qwen3.5-35B-A3B	91.6%	estimated ± 0.5 pp, low confidence
21	GPT-5.2	91.1%	estimated ± 0.5 pp, low confidence
22	GLM-5.3-Flash	90.4%	estimated ± 0.9 pp, low confidence
23	Gemma 4 12B	89.1%	estimated ± 0.5 pp, low confidence
24	Claude Opus 4.5	85.6%	estimated ± 0.5 pp, low confidence
25	Gemini 3.7 Flash	85.5%	estimated ± 0.9 pp, low confidence
26	Muse Spark 1.1	83.5%	estimated ± 0.9 pp, low confidence
27	Claude Sonnet 5	82.9%	estimated ± 0.9 pp, low confidence
28	Sakana Fugu-Ultra	77.6%	estimated ± 0.9 pp, low confidence
29	Muse Spark	77.4%	estimated ± 0.9 pp, low confidence
30	Sakana Fugu	76.9%	estimated ± 0.9 pp, low confidence
31	Gemini 3.5 Flash	76.8%	estimated ± 0.9 pp, low confidence
32	GPT-5.4	76.8%	estimated ± 0.9 pp, low confidence
33	Inkling	76.8%	estimated ± 0.9 pp, low confidence
34	Inkling-Small	76.8%	estimated ± 0.9 pp, low confidence
35	MiMo-V2.5	76.8%	estimated ± 0.9 pp, low confidence
36	Gemini 3.1 Pro	76.8%	estimated ± 0.9 pp, low confidence
37	Muse Glimmer 30B	76.8%	estimated ± 0.9 pp, low confidence
38	Qwen3.6-27B	76.8%	estimated ± 0.9 pp, low confidence
39	Claude Sonnet 4.6	76.8%	estimated ± 0.9 pp, low confidence
40	Command A+	76.8%	estimated ± 0.9 pp, low confidence
41	Gemini 3.1 Flash-Lite	76.8%	estimated ± 0.9 pp, low confidence
42	Grok 4.20	76.8%	estimated ± 0.9 pp, low confidence
43	Nemotron 3 Nano Omni 30B A3B	76.8%	estimated ± 0.9 pp, low confidence
44	Qwen3.6-35B-A3B	76.8%	estimated ± 0.9 pp, low confidence

MathVision w/ Python leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General