Trying to work out which Claude Code effort level you should actually be using? The quick answer is just below, and then we will go deeper — including the brand-new maxEffortLevel cap that shipped in version 2.1.267 on 9 September 2026.
Short answer
- Effort levels run
low→medium→high→xhigh→max; high is the default on most models (xhigh on Opus 4.7). - Run
/effortwith no arguments for an interactive slider;/effort autoclears your saved level. - Higher effort = deeper reasoning but more tokens;
maxcan overthink — the docs literally warn about diminishing returns. ultracodecombines xhigh reasoning with automatic workflow orchestration.- New in 2.1.267 (9 September 2026):
maxEffortLevelcaps effort on every provider, including Bedrock, Vertex and Foundry.
What Is the Claude Code Effort Level?
The Claude Code effort level controls adaptive reasoning — it lets the model decide whether, and how much, to think on each step based on task complexity. Per the official Claude Code documentation, adaptive reasoning “makes thinking optional on each step, so Claude can respond faster to routine prompts and reserve deeper thinking for steps that benefit from it”.
Practically, that means effort is your main dial for trading speed and token spend against intelligence. Low effort answers routine prompts fast and cheap; max effort burns more tokens reasoning deeply. The Fable models, Sonnet 5 and Opus 4.7+ always use adaptive reasoning, while Opus 4.6 and Sonnet 4.6 can revert to a fixed thinking budget with an environment variable.
Every Claude Code Effort Level Compared
| Level | When to use it (per the official docs) |
|---|---|
low |
Short, scoped, latency-sensitive tasks that are not intelligence-sensitive |
medium |
Cost-sensitive work that can trade off some intelligence |
high |
Default on most models — balances token usage and intelligence |
xhigh |
Deeper reasoning at higher token spend; default on Opus 4.7 |
max |
Demanding tasks — may show diminishing returns and overthinking, so test before adopting broadly |
ultracode |
A setting rather than a plain level: xhigh reasoning plus automatic dynamic workflow orchestration |
Model support matters too: Fable 5.1, Fable 5, Opus 5, Sonnet 5, Opus 4.8 and Opus 4.7 support all five levels, while Opus 4.6 and Sonnet 4.6 support low, medium, high and max only. Set an unsupported level and Claude Code quietly falls back to the highest supported level at or below it — xhigh runs as high on Opus 4.6, for example.
🔥 Want this set up without the guesswork? Dialling in your Claude Code effort level is exactly the kind of small tweak that decides whether AI content workflows are profitable or expensive. That is exactly the kind of thing we work through inside the AI Profit Boardroom — 3,700+ members, four live calls per week, daily tutorials, done-for-you templates and a 30-day roadmap.
Not sure where Claude Code fits in your SEO stack? Book a free SEO strategy session and we will map it out together.
How to Change It: /effort, Settings and Flags
You have five documented ways to set the effort level:
/effortwith no arguments opens an interactive arrow-key slider;/effort highsets a level directly;/effort autoclears your saved level for the active model.- Save behaviour: pressing Enter saves the level as that model’s default for later sessions; pressing
sapplies it to the current session only (v2.1.257+). - Settings:
"effortLevel": "high"globally, or per-model undermodelSettings— e.g. xhigh just forclaude-opus-5. - Environment variable:
CLAUDE_CODE_EFFORT_LEVEL=high. - Launch flag:
claude --effort high.
Resolution order, when several of these disagree: explicit choices (env var, flag, /effort) win, then the model’s default effort hold, then modelSettings or the global effortLevel, then the model’s own default.
maxEffortLevel: The New Cap in 2.1.267
Released 9 September 2026, per the official Claude Code changelog: the maxEffortLevel setting now “caps the effort level on every provider, including Bedrock, Vertex and Foundry”. This is the managed-settings control teams have wanted — an organisation can set "maxEffortLevel": "xhigh" and no one on the team can burn tokens on max, whichever provider their traffic routes through.
The cap is enforced in the UI too: levels above the cap are not offered in the /effort picker, and naming a higher level with --effort or /effort simply runs at the cap instead. If you manage a team’s Claude Code spend, this one setting is the story of the week.
Ultracode and ultrathink: The Two Odd Ones Out
Two names confuse everyone, so here is the clean split. Ultracode is a mode: xhigh per-message reasoning plus automatic dynamic workflow orchestration for substantive tasks. Enable it with /effort ultracode, claude --effort ultracode, the /model picker slider, or "ultracode": true in settings — note the persistent effortLevel setting and the environment variable do not accept it. Ultrathink is a one-off keyword: include the word ultrathink anywhere in a prompt and Claude Code requests deeper reasoning for that single turn without changing your session’s effort setting. Other phrases like “think hard” pass through as ordinary text.
The bottom line on the Claude Code effort level
Leave high as your daily default, save xhigh per-model for genuinely hard work, treat max as a tool you test rather than a setting you live on, and use the new maxEffortLevel cap if you are the one paying the team’s token bill. If you want to see how we wire Claude Code into actual revenue-generating SEO systems — not just settings tweaks — that is what the AI Profit Boardroom is for, or you can book a free SEO strategy session and get a plan for your specific site.
FAQ: Claude Code Effort Level
What is the default Claude Code effort level?
High is the default on most models, per the official Claude Code documentation. Opus 4.7 defaults to xhigh. If you have never touched /effort, you are almost certainly running on high.
Does a higher effort level cost more?
Generally yes — the docs describe xhigh as ‘deeper reasoning at higher token spend’, and lower levels as faster and cheaper for straightforward tasks. Effort trades tokens (and therefore cost and speed) for reasoning depth.
What is ultracode in Claude Code?
Ultracode is a Claude Code setting that combines xhigh reasoning effort with automatic dynamic workflow orchestration for substantive tasks. You enable it with /effort ultracode, claude –effort ultracode, or “ultracode”: true in settings — the plain effortLevel setting does not accept it.
Can I set a different effort level per model?
Yes. Claude Code saves per-model levels under modelSettings in user settings — for example giving claude-opus-5 xhigh while everything else stays on high. Setting a level interactively and pressing Enter saves it as that model’s default.
What happens if my model does not support the level I set?
Claude Code falls back to the highest supported level at or below the one you set. The docs give the example that xhigh runs as high on Opus 4.6.
Is max always the best effort level?
No. The official docs warn that max ‘may show diminishing returns and overthinking’ and recommend testing it before adopting it broadly. For most day-to-day work, high or xhigh is the sweet spot.
Related reading
- Claude Code /skill-doctor: prune the skills wasting your context
- Claude Code diff panel: see every edit live
- Claude Code self-hosted environments: the beta explained
Where to go next: Settings are only useful when they serve a system. Join the AI Profit Boardroom for the live calls, templates and the 30-day roadmap.
Rather start with a conversation about your site? Book a free SEO strategy session — no cost, no pressure, just a plan.
About the author
Julian Goldie is an SEO agency owner with 10+ years in SEO, 394K+ YouTube subscribers, a 100% job-success score on Upwork, 75K+ community members across his groups, and the author of a best-selling SEO book. He publishes daily AI SEO tutorials on YouTube, runs the AI Profit Boardroom community, and offers a free SEO strategy session if you want a personal growth plan for your site.
Last updated September 2026. This is the living guide to claude code effort level — it gets updated as the tools change.
