Report: GPT-6 Astra post-training unfinished, compute-to-performance scaling still unstable

Leaked reports and benchmark chart analyses suggest GPT-6 Astra's post-training is far from complete, with the model unable to reliably convert more inference compute into better results, indicating an unstable scaling curve.

2026-09-04 ~ 2026-09-04 · 2 related posts

• Astra's Post-Training Is Far From Done: More Compute Doesn't Monotonically Help Yet — teortaxesTex · 2026-09-04
• Leak claims OpenAI's Astra post-training unfinished, can't yet convert compute into gains — teortaxesTex · 2026-09-04