Claude Code effort levels: which one fits each task (cheat sheet)

Straight from Claude Code's model configuration docs.

1. low: quick exchanges you review: brainstorming, a first sketch, a rename.
2. medium: the default on Opus 5.5 and Sonnet 5.5. Day-to-day work with a clear scope, like building a feature.
3. high: when verification matters or edge cases are likely, like fixing a bug in existing code. Default on most other models.
4. xhigh: deeper reasoning at higher token spend.
5. max: hard problems Claude works through without you, like hunting security holes. The docs warn it can overthink: test before using it everywhere. Applies to the current session only.
6. ultracode: a setting, not a level. Claude plans a dynamic workflow for each substantive task, at whatever level you're on.

Tip: Picking the lowest level that does the job saves your usage limit for the jobs that need it.

Official docs →

Enter in the /effort slider, or typing a level after /effort, saves it as your default. Press s in the slider to apply it to this session only. You can also set effortLevel in settings, the CLAUDE_CODE_EFFORT_LEVEL variable, or launch with claude --effort high.