Inference
EN
Vals AI

SkillsBench

How important are skills for agents?

ModelsModels with a result.35
Top resultBest result on this benchmark.69.8%DeepSeek V4.1 Flash 1017
Top-3 spreadPoints from first to third.3.9 pts
UpdatedDate the source changed the results.29 Oct 2026

Result and cost

35 of 35 results

Results

35 results
#ModelResultThe score from the source.CostUS dollars to run one task.TimeTime to run one task.
01DeepSeek V4.1 Flash 1017OpenHands69.8% ±3.9$0.0854 min
02Grok 4.5OpenHands66.0% ±4.6$0.5782 min
03Gemini 3.7 FlashOpenHands65.9% ±4.6$1.803 min
04GPT-5.5 CodexCodex62.6% ±4.4$1.575 min
05GPT-5.5OpenHands62.2% ±4.5$2.547 min
06Claude Fable 5.1OpenHands61.6% ±4.8$4.5211 min
07GPT-5.6 LunaOpenHands60.4% ±4.8$0.2075 min
07Claude Opus 5OpenHands60.4% ±4.6$2.518 min
09Claude Opus 4.8OpenHands59.2% ±4.5$4.7716 min
09Muse Spark 1.1OpenHands59.2% ±4.4$0.4496 min
11GPT-5.6 TerraOpenHands58.9% ±4.5$1.757 min
12Gemini 3.8 FlashOpenHands58.0% ±4.4$2.344 min
13Grok 4.6OpenHands55.8% ±4.8$0.92113 min
14Qwen3.7 PlusOpenHands54.3% ±4.3$0.13712 min
15GPT-5.6 SolOpenHands54.1% ±4.7$4.1413 min
16DeepSeek V4 Pro 0813OpenHands53.8% ±4.5$0.37511 min
17Muse Spark 1.2OpenHands53.0% ±4.4$0.64913 min
18Gemini 3.5 FlashOpenHands52.7% ±4.5$1.257 min
19GPT-5.4OpenHands51.7% ±4.7$2.049 min
20MiniMax M3OpenHands51.5% ±4.5$0.50515 min
21DeepSeek V4 Pro 0423OpenHands51.3% ±4.6$0.3978 min
22DeepSeek V4 Flash 0731OpenHands50.7% ±4.4$0.1267 min
23Kimi K2.7 CodeOpenHands50.0% ±4.2$0.44313 min
24Claude Sonnet 4.6OpenHands49.1% ±4.6$1.3814 min
25GLM-5.3OpenHands47.5% ±4.5$0.71413 min
26Claude Sonnet 5OpenHands46.5% ±4.5$2.9015 min
27GLM-5.2OpenHands45.1% ±4.4$0.81120 min
28Qwen3.8 MaxOpenHands42.0% ±4.4$0.52516 min
29Grok 4.3OpenHands40.6% ±4.6$0.425 min
30GLM-5.3-FlashOpenHands40.2% ±4.4$0.07812 min
31Qwen3.8-27BOpenHands38.1% ±4.4$0.8187 min
32Inkling-SmallOpenHands33.6% ±3.7$0.14812 min
33Ling 3.0 FlashOpenHands27.9% ±3.8$0.0615 min
34InklingOpenHands26.1% ±3.9$0.82716 min
35Mercury 2.5OpenHands18.1% ±3.1$0.10149 s