Cohere Labs Studied How to Scale Multi-Agent Teams for Long-Horizon Coding
Cohere_Labs · x · 2026-10-15
- Cohere Labs announced new research on scaling multi-agent teams for long-horizon software engineering tasks, with an accompanying paper and blog post.
- The team built a multi-agent harness from scratch and ran experiments on challenging coding tasks: adding more coding agents helps on hard problems, but the benefit isn't free.
- The central question is how to organize the agent team — coordination structure and orchestration shape both performance and cost on long-horizon work.
Related event: Cohere Research: Where You Place Agents Matters More Than How Many(18 posts)→
More from coding & agent
- Doc2Harness benchmarks LLMs on harness engineering: GPT-6 Astra tops at 53.05%, still failing half — Yuyao Ge · 2026-10-15
- Long-horizon agent RL recipe: Muon optimizer with single-rollout async training boosts repo-level issue resolution — Kai Ruan · 2026-10-15
- Reality check: 10 self-distillation agent RL methods fail to consistently beat GRPO — zju · 2026-10-15
- Building a 'hypnotic piano groove' skill for AI music systems, modeled on Todd Terje — doodlestein · 2026-10-15
- Most Software Has No Access Layer Between the Agent and the App — That's the Problem — usehive · 2026-10-15
- Dev joke petition: forget a Codex reset, give us a Claude Code reset instead — weswinder · 2026-10-15