Update on Claude Code quality reports
Anthropic traced Claude Code quality complaints to three product changes, including a default-effort cut from high to medium, not to model degradation.
March 4: default reasoning effort lowered to cut latency, reverted April 7. March 26: a bug cleared older thinking every turn, fixed April 10. April 16: a verbosity system-prompt change hurt coding quality, reverted April 20. The API and inference layer were unaffected; all fixed by v2.1.116.
- Date
- Thursday, 23 April 2026
- Lab
- Anthropic
- Kind
- infra
- Access
- closed API
Anthropic says it never intentionally degrades models; the account is the company's own.
Sources
This record was checked against its sources on 6 October 2026. How we check