Usage limits
How to check your Claude usage (Pro, Max, and API — 2026)
Surfaces verified July 2026; vendor UI moves, so treat exact menu paths as a guide.
There is no single screen that shows everything. Consumer subscription usage, Claude Code's own counters, and API billing are three separate meters with three separate surfaces.
TLDR: For a quick answer inside a coding session, ask the Claude Code CLI. For plan-level consumption, check your account settings on the web app. For API spend, use the billing console — it is a different meter entirely and will not reflect subscription usage. None of the three forecasts when you will run out; that requires reading your own transcripts.
Short answer: there is no one screen. Claude usage lives on three separate meters — your subscription plan, the Claude Code CLI’s view of your current session, and API billing — and they answer different questions. Below is what each surface tells you, what it does not, and the question none of them answer.
The three meters
| Surface | What it tells you | What it will not tell you |
|---|---|---|
| Claude web app, account settings | Plan-level consumption against your subscription | Per-project or per-session breakdown |
| Claude Code CLI status | Where you stand in the current session and window | When you will run out at your current pace |
| API billing console | Spend and throughput on an API key | Anything about subscription limits |
| Local transcripts | Everything, per session, per project | Nothing — but you have to read them |
Surfaces as of July 2026. Vendor UI changes; the meters themselves have been stable.
In the Claude Code CLI
If you are already in a coding session, this is the fastest route. Claude Code exposes a status command that reports on your current session and window without leaving the terminal.
This is the right surface for “am I about to get cut off in the next hour”. It is the wrong surface for “how much of my week is left”, because it reports position rather than trajectory.
In the Claude web app
Plan-level consumption lives in your account settings on the web app. This is where you look to answer “how much of my subscription have I actually used”.
A caveat worth stating plainly: Anthropic has reorganised these surfaces more than once, which is why a lot of search results confidently describe menus that have since moved. Rather than following a click-path from a blog post, look under account or settings for the current usage view. If your Max tier is the question, our Claude Max limits page covers what the higher tiers change.
API usage is a different meter
If you are calling api.anthropic.com with an API key, none of the above applies. That is metered separately — on request and token throughput against a tier — and it surfaces in the billing console, not your subscription.
Mixing these up wastes real time. Upgrading your Claude plan will not raise an API rate limit, and a spent weekly cap will not be fixed by a higher API tier. We separate the two in Anthropic rate limits.
What counts toward your limits
Broadly: everything the model reads and writes on your behalf. That includes the parts you did not type — tool and command output pulled into context, files re-read across turns as the agent re-orients, and conversation history carried forward.
This is why usage often feels disconnected from effort. A short prompt in a large repository can cost far more than a long prompt in a small one. The mechanics of the two clocks are in Claude Code usage limits and the 5-hour window.
The question none of them answer
Every surface above is a rear-view mirror. They report where you have been, not where you are heading.
The forecast question — at my current pace, when do I hit the cap? — has to be computed from your usage history. Your local Claude Code transcripts hold that history, which is what the free audit CLI reads to produce a breakdown, and what Hypermile uses to project the cap and alert you before you hit it rather than after.
For the full field of tools that read those transcripts, including the free and open-source options, see Claude usage trackers compared — or download Hypermile.
Frequently asked questions
How do I check my Claude usage?
It depends which meter you mean. Inside a Claude Code session, the CLI's status command is the fastest answer. For plan-level consumption, look in your account settings on the Claude web app. For API key spend, use the billing console. These are separate meters and the numbers are not interchangeable.
Where is the Claude usage dashboard?
Anthropic has moved and reworked usage surfaces more than once, which is why so many search results point at menus that no longer exist. Rather than following a stale click-path, look under account or settings in the web app for the current usage view, and use the CLI for in-session numbers.
Does Claude Code show how much of my limit is left?
The CLI reports on your current session and window, which tells you where you stand right now. What it does not do is project forward — it will not tell you that at your current pace you will exhaust the weekly cap on Thursday. That projection has to be computed from your usage history.
Is API usage counted against my Pro or Max plan?
No. API usage billed against an API key is a separate meter from the consumer subscription's rolling window and weekly cap. Spending on one does not draw down the other, and hitting API rate limits is a different problem with a different fix.
Why does my usage look different in two places?
Almost always because you are reading two different meters — subscription consumption versus API spend — or because one surface reports the rolling window and the other reports the fixed weekly cycle. Confirm which clock a number refers to before comparing them.