Microsoft paper: Distilled skills replace reasoning, cutting tokens by 2.9-4.5x
rohanpaul_ai · x · 2026-08-15
A new Microsoft paper proposes reducing test-time reasoning costs by extracting recurring failure patterns from past agent trajectories and turning them into small markdown skills added to the system prompt of a non-reasoning model. Experiments show this recovers 55%-100%+ of the gap between non-reasoning and reasoning modes across 4 agent benchmarks, while using 2.9-4.5x fewer output tokens. On ALFWorld and τ²-retail, the skilled non-reasoning model even beat the reasoning mode. The distiller did not need reasoning traces; skills built only from cheap non-reasoning rollouts were competitive across all 4 domains. However, reasoning still won on telecom and SpreadsheetBench, where tasks have more instance-specific dependencies. Practical takeaway: distill repeated procedures once, reserve expensive reasoning for tasks needing fresh search.
More from coding & agent
- Coding agents keep opening 3 stacked PRs per chat, and merging them is a nightmare — holdenmatt · 2026-10-16
- Google-led paper uses sequential auctions to organize decentralized LLM agent workflows — _onionesque · 2026-10-16
- Since OpenAI's math release, GitHub contributors keep driving down the integer-multiplication kappa — xennygrimmato_ · 2026-10-16
- Pattern MCP makes coding agents check existing components before writing new code — DonR954 · 2026-10-16
- Zero-keyframe procedural spider walks walls and ceilings using pure raycasts — yshan2u · 2026-10-16
- OpenCode v1.19.2 keeps 1.x installs off the 2.x update path — opencode-agent[bot] · 2026-10-16