claude-code - 💡(How to fix) Fix Degraded reasoning quality and compute concerns — paying user reporting session failures [1 participants]

1. "Let us know" phrasing — The assistant said "let us know what happens" — a basic error for a reasoning model that should be aware it is a single AI instance, not a team.

TL;DR

The user should request confirmation from Anthropic that the claude-sonnet-4-6 model is receiving consistent and uncompromised compute resources to address the noticeable drop in quality.

Guidance

• The user should ask Anthropic to provide transparency about any changes to model performance or compute allocation that may have affected the quality of the claude-sonnet-4-6 model.
• Anthropic should verify whether the compute backing claude-sonnet-4-6 has changed recently, which could be causing the quality drop.
• The user should request a clear explanation of any changes made to the model or compute resources and when these changes were implemented.
• Anthropic should provide assurance that paying users, such as those on the Max plan, will receive consistent and uncompromised compute resources.

Notes

The issue lacks technical details about the model or compute resources, so it's difficult to provide a more specific solution. The user's request for transparency and confirmation from Anthropic is a reasonable step to address the quality drop.

Recommendation

Apply workaround: Request confirmation and transparency from Anthropic about the compute resources and model performance to ensure that paying users receive the expected quality. This is because the user has already experienced a noticeable drop in quality and needs assurance that the issue will be addressed.