Anthropic staff test how effort in Claude Code affects Fable 5.1 and Opus 5.5 output quality

Anthropic 员工实测 Claude Code 的 effort 档位如何影响 Fable 5.1 与 Opus 5.5 输出质量

Anthropic staff found in hands-on tests that raising effort in Claude Code adds verification and edge-case testing for Fable 5.1 and Opus 5.5, but usually cannot fix a wrong solution direction.

Why it matters: After testing, the staff adopted a workflow of implementing at low effort and verifying at high effort, splitting fast iteration and edge-case testing across different stages.