“It’d be like going from not even having o1, just having non-reasoning models, to Astra in a single year. So I do think things go faster.”
Books mentioned in OpenAI researcher on agent swarms & recursive self-improvement
From Dwarkesh Podcast · 2026-09-17