Astra reportedly uses looped transformers for better performance

rohanpaul_ai · x · 2026-09-02

Reports indicate that Astra uses "recurrent depth" or "looped transformers." Unlike standard models passing through a fixed stack once, a looped transformer runs the same information through the same layers multiple times before producing the next token. This approach helps improve performance but makes internal reasoning less readable.

Related event: OpenAI's Astra Reportedly Uses Recurrent Depth Architecture, Hiding Its Reasoning and Raising Safety Alarms(35 posts)→

Original post →

More from Models

Models channel →