Microsoft's Agensh: boss-free coding agent teams keep improving from 1 to 128 agents
rohanpaul_ai · x · 2026-10-05
A new Microsoft paper introduces Agensh, a self-organized multi-agent coding harness with no central orchestrator: each agent claims its own sub-task, builds and tests it, merges into a shared Git repo, and logs findings on a shared board.
On the 5 hardest of ProgramBench's 200 tasks (using GPT-5.6-sol high), scaling from 1 to 128 agents raised the mean test-pass rate from 19.31% to 28.78% (49% relative gain), improving at every step, and larger teams reached comparable scores sooner. On pandoc, scaling to 1,024 agents pushed the pass rate from 33.89% to 55.06%.
Popular multi-agent coding tools funnel all work through a single lead agent with limited bandwidth. The paper used a single model throughout and doesn't report the cost of large teams.
Related event: Microsoft's Agensh Scales Decentralized Coding Agents to 1,024(2 posts)→
More from coding & agent
- Why Notion is best positioned to win multiplayer humans-plus-agents knowledge work — ivanhzhao · 2026-10-05
- Researcher hijacks Copilot in SQL Server Management Studio, escalating from SELECT to SYSADMIN (CVE-2026-65669) — wunderwuzzi23 · 2026-10-05
- User discovers Codex's dot can now drive Chrome on a local computer — TheMoonMidas · 2026-10-05
- Stop wasting your AI subscription compute: use the smartest model only when it matters — TheChuckTone · 2026-10-05
- Free AI Engineering course covers foundation models, evals, prompt engineering and security — Negative_War_65 · 2026-10-05
- ₹40,000 web dev bootcamp in 2026? Claude and Cursor already write full-stack apps — _jaydeepkarale · 2026-10-05