Agent swarms collaborate for 10,000 hours in benchmark test

The World of SWECraft project ran frontier-model agent swarms for 10,000 hours collaborating on real software in a simulated agent-only GitHub, with Gemini-3.8-Flash leading (23% speedup) and gpt-6-astra ranking last.

2026-10-23 ~ 2026-10-23 · 2 related posts

• Agent swarms collaborate 10K hours in simulated GitHubs: Gemini-3.8-flash tops, gpt-6-astra worst — RulinShao · 2026-10-23
• Agent Swarms Tested for 10K Hours: Gemini-3.8-Flash Leads SWECraft With 23% Speedup, GPT-6-Astra Worst — mrdrozdov · 2026-10-23