astra's benchmark numbers are the least interesting part.
searchable codex history and stop/ask behavior are closer to the failures i actually hit.
the useful agent is not the one that never asks. it knows what can continue safely without me.
openai.com/index/gpt-6-...
openai.com
GPT-6 Astra: A new generation of intelligence
Introducing GPT-6 Astra, our most intelligent and aligned model yet, with state-of-the-art capabilities across computer use, coding, cybersecurity, and science.