User finds GPT-6 Astra weak at orchestration, would use it as expert call-in only

koltregaskes · x · 2026-09-09

A user reports GPT-6 Astra performs poorly for their work, particularly at orchestration. They'll keep testing briefly but expect to go back to GPT-5.6 Sol, positioning Astra instead as an expert call-in within a Sol session.

Original post →

More from Models

Models channel →