About ClaudeEffort
An independent calculator for Claude effort levels and API cost, built and maintained by one developer.
ClaudeEffort is an independent, single-purpose tool: it tells you which Claude effort level a task needs, and what that choice costs. It was built and is maintained by Charles, a developer who spent too long trying to reconcile an LLM bill against the parameters that produced it.
The site is not affiliated with, endorsed by, or sponsored by Anthropic or OpenAI. It has no accounts, no signup and no server-side storage of anything you type.
Why this site exists
The effort parameter is the largest lever on a Claude bill that most teams never touch, and it is also the one that is hardest to reason about from the outside. Two things make it awkward:
- Effort scales every output token, not just the reasoning ones — answer text, tool calls and their arguments included. A description like "how much it thinks" leads people to the wrong mental model.
- The multiple on your output tokens is never the multiple on your bill, because input cost is a fixed floor. At
lowandmediumthe input side is close to half of what you pay.
Neither of those is obvious from a pricing page, and neither shows up until someone looks at a real invoice. So the site does two jobs: it recommends a level for a task, and it prints the arithmetic behind that recommendation instead of hiding it.
What makes it different from a price table
A price table tells you the rate per million tokens. It does not tell you which of five effort levels your classifier should be running at, or whether raising effort on a failing task is cheaper than switching to a stronger model.
Every number on this site is traceable to one of two places:
- Vendor-published rates, read from the vendor's own pricing documentation and dated in the footer.
- Planning assumptions, which are labelled as assumptions wherever they appear and which you can override in the calculator.
The distinction is enforced everywhere, including in the articles. You can read the full breakdown on the methodology page.
How it is funded
The site carries advertising. That is the whole business model — there is no paid tier, no affiliate arrangement with Anthropic or OpenAI, and no sponsored content. Advertising has no influence on which effort level the tool recommends for a given task; those recommendations follow from the cost arithmetic, and you can check the arithmetic yourself on any page.
Editorial rules
- Prices are re-verified against the vendor's official pricing documentation, and the verification date is published in the footer of every page rather than buried in a changelog.
- Where a figure is an assumption rather than a published rate, it is labelled as an assumption in the same paragraph it appears in.
- Nothing is published that cannot be recomputed from the numbers shown. If a table on this site disagrees with the calculator, that is a bug, and we want to hear about it.
- Corrections are made to the page rather than to a separate errata note, and the "updated" date on an article changes when its numbers change.
Getting in touch
Corrections, disagreements about methodology and questions all go to the same place. See the contact page for the address and what to expect.
Where to start
- The calculator, for a per-call and monthly number on your own workload
- Effort levels explained, for what each of the five levels does
- Guides, for the longer write-ups on billing, caching and cost control