First Recursive Self-Improvement Applied to Frontier Pre-training, Open-Sourced 30B Model Beats Qwen3
SII-GAIR and Xinxu AI published 'No Second Run,' claiming the first application of recursive self-improvement (RSI) to full from-scratch pre-training, producing the open-sourced 801-30B-A3B model in about three months, which outperforms Qwen3.
2026-11-12 ~ 2026-11-13 · 2 related posts
- First RSI on a Frontier Pre-training Run: Open 801-30B Beats Qwen3-30B — dair_ai · 2026-11-12
- First RSI applied to full pre-training: 801-30B-A3B trained from scratch in 3 months on 11.8T tokens — Xudong07452910 · 2026-11-13