OpenAI's next model thinks in loops that leave fewer traces.
TechCrunch reported on 2 September that Astra uses a technique called recurrent depth. The model repeats a query internally instead of writing every reasoning step out in text. OpenAI says it kept the loops shallow, so Astra's visible reasoning should stay readable.
Why it mattersRedwood Research warns that deeper loops would end chain of thought monitoring altogether. Reading a model's working notes is one of the few safety checks that still works.
Read in the app — free, in 9 languages
Related stories
Perplexity cites 215,128 buying guides made by three linked websites.2026-09-02OpenAI's next model scored 100 percent on an exploit-writing test.
2026-09-02One developer beat many chatbots on a reasoning test for 67 cents.
2026-09-02Anthropic found faults in over 10 percent of its own training tasks.
2026-09-02The best AI agent finished 30% of real lab work
2026-08-29