Early users say Astra doesn't wait on consequential decisions as OpenAI's blog claims

ChrisGPT · x · 2026-09-05

The author finds GPT-6 Astra smart and token-efficient but disputes one OpenAI blog claim: that the model "waits for your input on consequential decisions." Two people with early access said it doesn't do this at all, raising the question of whether it depends on an undisclosed Codex configuration.

Original post →

More from Models

Models channel →