# Allspark: Weak to Strong Transfer via Alternating Chain of Thought

> Source: <https://aiflash.com/news/128753/>
> Published: 2026-09-29 19:00:07+00:00

Recent progress in frontier models has renewed interest in large-scale reinforcement learning (RL), but the cost of generating large-model rollouts makes even testing RL recipes expensive. We ask whether reasoning improvements learned by a small, weak model can benefit a larger, stronger model witho