SkillOS: Learning Skill Curation for Self-Evolving Agents
Open paper
Benchmark
Model
Rank
Results
mathematical-reasoning-on-aime-2025
SkillOS (Qwen3-8B curator) with Gemini-2.5-Pro executor
–
Accuracy: 86.7
mathematical-reasoning-on-aime24
SkillOS (Qwen3-8B curator) with Gemini-2.5-Pro executor
–
Acc: 92.2
Rank counts only results with a code link.