RSIAgent: Training-Free Multi-Agent Framework Beats GPT-6 via Accumulated Experience

Aether AI, with UC San Diego and UIC researchers, released RSIAgent, a training-free multi-agent framework where agents pose tasks and build memory to achieve recursive self-improvement, outperforming GPT-6.

2026-09-21 ~ 2026-09-23 · 2 related posts

• RSIAgent: Training-Free Agent Self-Improvement Beats GPT-6 on OSWorld — alex_verem · 2026-09-21
• RSIAgent beats GPT-6 Astra on OSWorld 2.0 without updating model weights — jiqizhixin · 2026-09-23