Claude Code effort levels explained
What each Claude Code effort level does, low to max, the default for each model, how to set it with /effort, settings or env, and which level fits a task.
Effort sets how much Claude thinks before it answers. Claude Code has five levels: low, medium, high, xhigh and max. Opus 5.5, Sonnet 5.5 and Haiku 5.5 default to medium. Thinking tokens bill as output tokens, so a lower effort costs less, and a higher one helps on hard problems. Set it with /effort.
Which effort level should I use?
| Level | Fits |
|---|---|
| low | Quick exchanges you review yourself: brainstorming, a first sketch, a rename |
| medium | Day-to-day work with a clear scope, such as a new feature |
| high | Work where verification matters, such as fixing a bug in existing code |
| xhigh | Deeper reasoning, at higher token spend |
| max | Hard problems Claude works through alone, such as a security review. It can overthink, so test it first |
Paraphrased from Anthropic's model configuration docs, read on 2026-11-05.
What is the default effort?
Opus 5.5, Sonnet 5.5 and Haiku 5.5 default to medium. Most other models that support effort default to high, and Opus 4.7 defaults to xhigh. If you set a level a model does not support, Claude Code uses the highest supported level below it. For example, xhigh runs as high on Opus 4.6.
How do I set the effort level?
- In a session:
/effortopens a slider./effort highsets it directly, and/effort autoclears your saved level. - For one session:
claude --effort high. - Everywhere: the
CLAUDE_CODE_EFFORT_LEVELenvironment variable. It wins over every other setting. - As a saved default:
effortLevelin settings, or a per-model level inmodelSettings. Settings do not acceptmax. - For a skill or subagent:
effortin its frontmatter.
Why not set one level and forget it?
Because prompts differ. A quick question at high wastes thinking tokens, and a tricky bug at low wastes your time. Most people never change the level, because doing it by hand on every prompt is tedious.
AutoMode picks the effort for each prompt, and the model too. In our benchmark of mixed sessions it cost 51% less than Claude Code on Opus, with every step passing. See how it chooses, or estimate your own saving.
Install AutoMode
Paste these into Claude Code, one at a time, then run /reload-plugins. It is free, needs no key and no server, and starts on its own.
/plugin marketplace add mralexivy/AutoMode
/plugin install automode@automodeLast updated on
How to reduce Claude Code costs
Practical ways to cut Claude Code token use and spend: match the model to the job, lower effort, clear context, keep the cache warm, and route each prompt.
Haiku vs Sonnet vs Opus in Claude Code
Which Claude model to use in Claude Code for questions, everyday coding and hard reasoning, with the aliases, the defaults, and measured cost and pass rates.