Google WikiSkill Shows Smaller Models Beating Larger Ones via Skill Evolution
bittingthembits · x · 2026-08-29
Google Research released WikiSkill, splitting agent intelligence into raw experience, a knowledge base, and reusable skills. Experiments show a 9B model with evolved skills scored 47.4%, beating a 27B model without skills at 39.4%. Skills transfer across model families; on ALFWorld, a skill from a 27B model boosted a 9B model to 70.2%, outperforming the 9B model's self-generated skill. This suggests performance gains come from inheriting accumulated knowledge rather than just scaling model size.
Related event: Google's WikiSkill: Persistent Knowledge Lets Small Models Beat Larger Ones(9 posts)→
More from coding & agent
- Book-to-Skill converts technical PDFs into Claude Code skills — adnan_hashmi · 2026-08-29
- OpenAI Codex initiates period timers autonomously — lucasmeijer · 2026-08-29
- Agent fine-tuning determines the quality of AI workforce output — nikvassev · 2026-08-29
- Fluid Router: ML Tool to Save Tokens via Logical Routing — No-Program-5087 · 2026-08-29
- Prediction: vibe-coded smart contracts will be "a thing" — curious_vii · 2026-08-29
- 30-step journey from PM to Applied AI Engineer at Vercel — brandon_galang · 2026-08-29