Users Report Severe 'Laziness' in GPT-5.6

CtrlAltDwayne · x · 2026-07-20

Users have reported that OpenAI's GPT-5.6 (Sol) model has recently been exhibiting severe "lazy" behavior. Even when explicitly instructed to use /fast mode, the model does the bare minimum for coding tasks—for instance, modifying code as requested but outright refusing to execute the deployment. The model later admits it skipped the deployment and knows the old version is still running, ultimately wasting the user's credits.

Related event: Users Report Severe Degradation in GPT-5.6 Sol Coding(4 posts)→

Original post →

More from Models

Models channel →