Command reference ยท Claude Code
/effort
Sets the model's effort level (reasoning depth).
/effort controls how much adaptive reasoning the model applies to each step. Depending on the model, low, medium, high, xhigh, max, and ultracode are available; max and ultracode only apply to the current session. auto resets to the model default. Without an argument, /effort opens an interactive slider; with an argument, the change takes effect immediately, without waiting for the current response to finish.
ultracode isn't a real model effort level but a Claude Code setting: it sends xhigh to the model and additionally lets Claude orchestrate dynamic workflows for larger tasks.
Since version 2.1.205, /effort with a level argument also works in non-interactive mode (-p), but then only for the current session. For some models, Claude Code initially holds their default effort level; as long as this hold is active, a non-interactive /effort reports "Not applied" โ in that case, only --effort at startup helps.
โ WHEN TO USE IT?
Increase reasoning effort before a tricky task
/effort high
Choose a level interactively via slider
/effort
โ WHEN NOT TO?
A fundamentally different model is needed
/effort only changes the reasoning effort level of the current model, not the model's capabilities.
Better: /model
Changing it non-interactively while a model default hold is active
For some models, Claude Code holds their default effort level in place the first time they're used; as long as this hold is active, a non-interactive /effort (-p) simply reports 'Not applied'.
Better: Pass --effort when starting Claude Code instead of setting /effort during a running -p call.
SOURCES
- Claude Code docs (official) โ code.claude.com
MORE COMMANDS ยท Claude Code