Agent fleet team on topping ClickBench: infrastructure is the bottleneck, engine to be open sourced
jiayq · x · 2026-10-15
Follow-up from jiayq's team on their agent fleet topping ClickBench (1.2× ahead of the next engine): as agents get more capable, surrounding infrastructure becomes the main limit on iteration speed, and letting the fleet work on its own methodology paid off. They stripped out anything that only helped this benchmark to avoid "AI feature creeps" — reviewer agents removed special snowflake optimizations — and their entry also ranks first on a concurrent QPS measure the fleet never targeted. The engine will be open sourced.
Related event: jiayq's Autonomous Agent Fleet Tops ClickBench, Engine to Be Open-Sourced(6 posts)→
More from coding & agent
- Open-Agent-Email: free open-source alternative to AgentMail for AI agents — matchaman11 · 2026-10-15
- AWS Gen AI Weekly Redesign: One-Page Exec Summary, AgentCore Memory Goes Usage-Based — DigitalColmer · 2026-10-15
- Nous Research Launches 50+ Connectors for Hermes Agent, No Account Required — NousResearch · 2026-10-15
- Dev group chat: skip runtime LLM web search, index products via API instead — vxnuaj · 2026-10-15
- Tev-Gym: 1,170 dense-reward RL environments push long-horizon agentic training — RulinShao · 2026-10-15
- Microsoft's AutoEnvScaling lets agents build their own RL environments, lifting Qwen3 Terminal-Bench from 41.1 to 53.3 — omarsar0 · 2026-10-15