Agentic · tools

SkillsBench leaderboard

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

As of 2026-10-07, the highest measured score on SkillsBench is 70.2% by Qwen3.8 Max. 10 more models have estimated scores, calibrated from the benchmarks they were measured on.

Measured scores: benchlm.ai.

#	Model	Score	Source
1	Qwen3.8 Max	70.2%	measured
2	Ling 3.1 Flash	68.7%	measured
3	Atria Dawn Preview	66.4%	measured
4	Hy4 preview	62.9%	measured
5	Kimi K2.6	61.3%	estimated ± 5.6 pp, low confidence
6	Ornith-1.5-397B	61.3%	estimated ± 5.6 pp, low confidence
7	dots3-note Preview	52.8%	measured
8	Claude Opus 4.5	51.2%	estimated ± 5.6 pp, low confidence
9	Qwen3.6 Plus	46.4%	estimated ± 5.6 pp, low confidence
10	Qwen3.5 397B	45.7%	estimated ± 5.6 pp, low confidence
11	Ling 3.0 Flash	44.8%	measured
12	Muse Glimmer 30B	44.3%	measured
13	Kimi K2.5	42.8%	estimated ± 5.6 pp, low confidence
14	GLM-5	36.3%	estimated ± 5.6 pp, low confidence
15	Ornith-1.5-35B-A3B	32.0%	estimated ± 5.6 pp, low confidence
16	Qwen3.6-35B-A3B	18.0%	estimated ± 5.6 pp, low confidence
17	Ornith-1.5-9B	17.1%	estimated ± 5.6 pp, low confidence

SkillsBench leaderboard

All leaderboards

Agentic · terminal

Agentic · tools

Coding

Math

Knowledge & reasoning

Instruction following

Multilingual

Vision & documents

Long context

General