Microsoft's FOCUS compresses agent context at test time: 48% less context, +8.9 points success
dair_ai · x · 2026-10-02
A new Microsoft paper introduces FOCUS, a training-free method that compresses agent context at test time by identifying which past interactions the agent's next decisions actually depend on and dropping the rest. It works as a standalone layer in front of closed-API models with no fine-tuning. On tool-calling, QA, web and multi-turn dialogue benchmarks it cuts peak context by up to 48% and improves task success by up to 8.9 points versus running on full history.
More from coding & agent
- Dev builds SCP Foundation survival game in Godot with Claude Opus 5.5 Max — imjustnewatai · 2026-10-02
- MockAgent: real-time AJV validation to catch agent tool-call hallucinations and retry loops — Zealousideal-Room775 · 2026-10-02
- Deedy's AI video workflow part 2: Claude Code, OpenRouter and emotional TTS setup — FinanceYF5 · 2026-10-02
- Dev turns personal bookmark library into an agent-accessible context source for Claude Code — dheeraj_iosdev · 2026-10-02
- Six Coding Agents, One Repo: Isolated Runs All Broke, Chatting Agents All Passed — jokiruiz · 2026-10-02
- Dev's AI Agent Astra Generates Game Assets via Pure Procedural Code, Not Imagen — Dimillian · 2026-10-02