Vision & documents

BabyVision leaderboard

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

As of 2026-10-07, the highest measured score on BabyVision is 82.0% by Qwen3.8 Max. 14 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Qwen3.8 Max	82.0%	measured
2	Kimi K3	79.9%	estimated ± 4.3 pp, medium confidence
3	Muse Spark 1.1	76.3%	measured
4	Seed 2.1 Pro	73.7%	measured
5	Qwen3.8-Omni-Flash	69.7%	estimated ± 4.3 pp, medium confidence
6	Qwen3.8-27B	65.7%	measured
7	Qwen3.8-Flash-Next	64.7%	estimated ± 4.3 pp, medium confidence
8	Qwen3.7 Plus	63.5%	estimated ± 4.3 pp, medium confidence
9	Seed 2.1 Turbo	62.9%	measured
10	Qwen3.5 397B	56.5%	estimated ± 4.3 pp, medium confidence
11	Qwen3.6 Plus	54.0%	estimated ± 4.3 pp, medium confidence
12	GLM-5.3-Flash	53.4%	measured
13	Kimi K2.6	51.6%	estimated ± 4.3 pp, low confidence
14	dots3-note Preview	50.0%	measured
15	Gemini 3 Pro	48.3%	estimated ± 4.3 pp, low confidence
16	Qwen3.5-122B-A10B	46.6%	estimated ± 4.3 pp, low confidence
17	Qwen3.5-27B	45.8%	estimated ± 4.3 pp, low confidence
18	Qwen3.5-35B-A3B	37.2%	estimated ± 4.3 pp, low confidence
19	GPT-5.2	33.5%	estimated ± 4.3 pp, low confidence
20	Gemma 4 12B	19.9%	estimated ± 4.3 pp, low confidence
21	Claude Opus 4.5	0.0%	estimated ± 4.3 pp, low confidence

BabyVision leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General