Hidden Features

Extended thinking

Let Claude reason at length before answering — and know when it helps.

Knowledge cards
Knowledge card
Give Thinking Room for Math, Logic, and Constraints
Turning extended thinking on for the class of problems that reward working through steps: multi-step arithmetic, logic and constraint puzzles, unit conversions, anything where a single wrong intermediate step silently poisons the final answer. These are the tasks where 'show the work' actually changes the result.
L3 · Workflows
Best practice
Read the Reasoning, Not Just the Answer
A habit of actually reading Claude's extended-thinking trace — not just the final answer — to spot where its reasoning took a wrong turn: a misread requirement, a bad assumption, a step skipped. The visible reasoning is a debugging surface for the answer.
L3 · Workflows
Best practice
Tell Thinking What a Good Answer Must Satisfy
Giving Claude an explicit definition of what a correct answer must satisfy — the constraints, the acceptance criteria, the must-haves — before it starts thinking, so its reasoning has a target to steer toward and check itself against, rather than optimising for 'a reasonable-sounding answer'.
L3 · Workflows
Best practice
Save Extended Thinking for Genuinely Hard Problems
A judgement call about when to turn on extended thinking — the mode where Claude reasons at length before answering. Reach for it on problems that genuinely need multi-step reasoning and where a wrong answer is costly; skip it on simple or easily-checked tasks where it just adds latency.
L3 · Workflows
Claude extended thinking

Let it reason before it answers

Extended thinking lets Claude work through a problem at length — visibly reasoning — before committing to a final answer. On hard problems, that deliberation genuinely improves the result: it catches its own missteps, considers alternatives, and arrives at a conclusion it can justify rather than a first impression. Seeing the reasoning also lets you check the path, not just the destination, which matters when the answer needs to be right. It is Claude thinking out loud instead of answering on reflex.

Know when it helps

Extended thinking is not free — it takes longer — so the skill is knowing when the depth is worth it. Turn it on for multi-step logic, tricky analysis, planning, and problems where a plausible-but-wrong answer is costly. For quick, simple tasks it is unnecessary overhead. The same judgement that guides model choice applies here: reach for more thinking when the problem earns it, and keep the fast path for the everyday work that does not.

Frequently asked
What is Claude's extended thinking?

A mode where Claude reasons at length, visibly, before giving its final answer — catching missteps and weighing alternatives, which improves results on genuinely hard problems.

When should I use extended thinking?

For multi-step logic, tricky analysis, and planning where accuracy matters. For quick, simple tasks it is unnecessary overhead, so reserve it for problems that earn the extra time.

Does extended thinking always give better answers?

Not for simple tasks, where it just adds latency. Its benefit shows on hard problems where deliberation catches errors a fast answer would miss.