SkillsBench

Agent tasks with and without packaged skills; overall score. Run by Vals AI.

Results

#	Model	Provider	Result	Date
1	DeepSeek V4.1 Flash High	DeepSeek	69.8%	2026-10-29
2	Grok 4.5 High	xAI	66.0%	2026-10-29
3	Gemini 3.7 Flash High	Google	65.9%	2026-10-29
4	GPT-5.5 Extra High	OpenAI	62.2%	2026-10-29
5	Claude Fable 5.1 Max	Anthropic	61.6%	2026-10-29
6	GPT-5.6 Luna Max	OpenAI	60.4%	2026-10-29
7	Claude Opus 5 Max	Anthropic	60.4%	2026-10-29
8	Claude Opus 4.8 Max	Anthropic	59.2%	2026-10-29
9	Muse Spark 1.1 Extra High	Meta	59.2%	2026-10-29
10	GPT-5.6 Terra Max	OpenAI	58.9%	2026-10-29
11	Gemini 3.8 Flash High	Google	58.0%	2026-10-29
12	Grok 4.6 High	xAI	55.8%	2026-10-29
13	Qwen3.7 Plus	Alibaba	54.3%	2026-10-29
14	GPT-5.6 Sol Max	OpenAI	54.1%	2026-10-29
15	DeepSeek V4 Pro 0813 Max	DeepSeek	53.8%	2026-10-29
16	Muse Spark 1.2 Extra High	Meta	53.0%	2026-10-29
17	Gemini 3.5 Flash High	Google	52.7%	2026-10-29
18	GPT-5.4 Extra High	OpenAI	51.7%	2026-10-29
19	MiniMax-M3	MiniMax	51.5%	2026-10-29
20	DeepSeek V4 Pro Max	DeepSeek	51.3%	2026-10-29
21	DeepSeek V4 Flash 0731 High	DeepSeek	50.7%	2026-10-29
22	Kimi K2.7 Code	Moonshot AI	50.0%	2026-10-29
23	Claude Sonnet 4.6 Max	Anthropic	49.1%	2026-10-29
24	GLM-5.3 Max	Zhipu AI	47.5%	2026-10-29
25	Claude Sonnet 5 Max	Anthropic	46.5%	2026-10-29
26	GLM-5.2 Max	Zhipu AI	45.1%	2026-10-29
27	Qwen3.8 Max	Alibaba	42.0%	2026-10-29
28	Grok 4.3 High	xAI	40.6%	2026-10-29
29	GLM-5.3-Flash Max	Zhipu AI	40.2%	2026-10-29
30	Qwen3.8 27B Extra High	Alibaba	38.1%	2026-10-29
31	Inkling Small	Thinking Machines	33.6%	2026-10-29
32	Inkling	Thinking Machines	26.1%	2026-10-29
33	Mercury 2.5 High	Inception	18.1%	2026-10-29