If you use both the Claude app and Claude Code, you’ve seen two different names for what sounds like the same thing: Extended Thinking in the app, and /effort in the terminal. They are the same underlying idea, just surfaced differently.
The one-line version
- Extended Thinking is the Claude app’s toggle for deep reasoning before answering.
/effortis Claude Code’s dial for the same thing — how much thinking Claude does before it acts.
Extended Thinking — the app toggle
In the Claude web or desktop app, you can turn on Extended Thinking for a conversation. When it’s on, Claude works through the problem step-by-step before giving you the answer — visible as a “thinking” phase.
It’s a binary choice in the app: on or off.
Best for: hard math, logic puzzles, careful planning, anything where a quick answer tends to be wrong.
Effort — the terminal dial
In Claude Code, you control the same capability with /effort:
/effort xhigh
But instead of on/off, you get a spectrum:
low— light touch, quick answers. Good for simple tasks.high— the default. Right for most work.xhigh— maximum reasoning depth. For the hard stuff.
There’s also /fast, which trades a bit of cost for noticeably faster output — the other direction on the speed/depth spectrum.
Side by side
| Extended Thinking (app) | Effort (Claude Code) | |
|---|---|---|
| Surface | Claude app (web/desktop) | Claude Code (terminal) |
| Control | On / off | Spectrum: low → high → xhigh |
| What it does | More reasoning before answering | More reasoning before acting |
| Default | Off (you opt in) | High (already engaged) |
| Best for | Hard questions in conversation | Hard coding/engineering tasks |
| Speed cost | Slower when on | Slower at xhigh, faster at low |
The practical advice
The rule is the same in both surfaces: match the thinking to the difficulty.
- Quick question, simple task → default effort / Extended Thinking off
- Hard problem, subtle bug, complex plan →
/effort xhigh/ Extended Thinking on - Batch of easy tasks →
/effort lowor/fast— don’t overthink the trivial
The mistake people make is leaving it at the extremes. Maximum thinking on every task wastes time. Minimum thinking on hard tasks wastes attempts. The skill is noticing when a problem deserves more thought — and giving it.