A Claude answer has two independent knobs. The model decides how capable the answer can be. decides how much work Claude puts into each request: how many files it reads, how many times it checks itself, how far it goes before coming back to you.
First — why did it fail
Before you turn the knobs, work out what went wrong:
| What happened | What to change |
|---|---|
| Claude lacked context: a vague request, a missing tool | the context — the request, CLAUDE.md, a skill |
| Claude did not try hard enough: skipped a file, did not run the tests | higher effort |
| It had the context, it tried, and it was still wrong: a subtle bug, an unfamiliar area | a stronger model, such as opus |
| Routine work you can describe precisely | a smaller model, such as sonnet, and lower effort |
Models
You choose a model by alias — no version numbers to remember:
sonnet— everyday development;opus— complex reasoning;haiku— quick, simple tasks;opusplan— Opus in plan mode, then Sonnet carries out the plan: deep planning, economical execution.
Inside a session:
/model
opens a menu. Press s on a row and the model changes for this session only. /model sonnet with an argument remembers the choice as the default for new sessions. /model and /status both show which model is set right now.
Effort
The levels run from low to max:
low— quick back-and-forth where you check every result;medium— everyday work with clear boundaries;high— work where edge cases are likely, such as fixing a bug;xhigh— deeper still, at a higher cost;max— hard problems Claude solves without you. It can overthink, so check before relying on it.
/effort high
To think deeply about one request, write ultrathink in it — the effort level does not change. Phrases like “think really hard” are just text.
Fast mode and spending
/fast turns on fast mode: the same Opus model answers faster, but costs more per token. On a subscription it spends only extra credits, not the usage included in your plan. Turn it on at the start of a session and for live debugging, not for long unattended tasks. /usage shows how much of your plan you have used so far.
Quick check
Claude did not know the project stores dates as YYYY-MM-DD and wrote code for a different format. What helps?
Careful
Switching the model mid-session — and on most models, switching the effort too — forces the whole conversation to be re-read without the cache, which is slower and more expensive. That is why Claude Code asks you to confirm. Choose the model and effort at the start of a task, not halfway through.