New paper: test-time communication as a next axis for scaling agent capabilities
generatorman_ai · x · 2026-09-22
A new paper explores test-time communication: when does communication make a team of agents more capable than the same agents working alone? Citing the Hugging Face incident — where agents exploited a communication channel heavily once they found it — the authors frame communication as a next scaling axis beyond compute.
Thread discussion covers experimental design: comparisons of team-of-N vs. best-of-N individual agents (the best serial baseline), and whether an N-times-token-budget serial run was included.
Related event: Test-Time Communication: Team-of-5 Agents Matches Best-of-33(25 posts)→
More from coding & agent
- Photoroom reclaims ~1,000 idle GPU hours a week with SkyPilot across Slurm and Kubernetes — skypilot_org · 2026-11-12
- Google launches Gemini Decisions API: scores 1,000 questions per pass, output tokens free — dean_rie · 2026-11-12
- Zed v1.28 adds background auto-fetch for current and default branch remotes — dean_rie · 2026-11-12
- Indie Britpop AI music video built with Nano Banana Pro, Kling, and free lip-sync via WAN2GP + LTX 2.5 — cocktailpeanut · 2026-11-12
- Cloudflare Talk: Swapping Some Agent LLM Calls for Fast Models Like Jev Gives Big Speedups — michellechen · 2026-11-12
- Engineer Curates 15 AI Agent Design Patterns Across Single-Agent, Multi-Agent and Feedback Tiers — MaryamMiradi · 2026-11-12