Users report GPT-5.6 Sol Max making repeated 'stupid' errors, speculating Astra launch is imminent

CtrlAltDwayne · x · 2026-09-03

A user reports GPT-5.6 Sol Max reasoning repeatedly making basic mistakes across multiple tasks: "One task, maybe a fluke. But multiple tasks? That's not a fluke." They speculate the quality degradation could signal that a new model called Astra is imminent — OpenAI has historically rerouted/downgraded existing models before launches — "otherwise we have a serious problem." Single-user observation, unverified.

Related event: Users Report Sharp Quality Drop in GPT-5.6 Max Reasoning(2 posts)→

Original post →

More from Models

Models channel →