selan.aiGuide · Claude Code limits
What do you do when Claude Code says rate limit reached?
On a Claude subscription, Claude Code stops when you fill one of two windows: a session limit that resets every five hours, or a weekly limit that resets at a fixed time. The message names the limit and its reset time. You can wait for it, buy usage credits at API rates, move to a bigger plan, or switch to an API key.
What are Claude Code's usage limits?
Pro, Max, Team and seat-based Enterprise plans have two: a session limit that resets every five hours, and a weekly limit across all models that resets once a week. Every request counts against both.
Claude Code, claude.ai and Claude Desktop share one allowance.
A long chat in Claude Desktop leaves less for Claude Code. How much fits in a window depends on the conversation's length, the model and the effort level. Anthropic publishes no fixed message count.
When does the Claude Code usage limit reset?
The session limit resets every five hours. The weekly limit resets at a fixed day and time assigned to your account, whenever you start using Claude. One heavy burst, such as a large fan-out of agents, can use up the weekly allowance before the session window resets.
/usage in Claude Code shows both windows and when each resets, and so does
Settings › Usage on claude.ai.
What does the rate limit message look like?
You've hit your session limit · resets 3:45pm
You've hit your weekly limit · resets Mon 12:00am
You've hit your Opus limit · resets 3:45pm
Claude Code blocks further requests until the time in the message. It warns you earlier,
with a line such as You've used 85% of your session limit.
From version 2.1.234, an interactive subscription session waits with the task open and carries on after the reset.
The bare API Error: Rate limit reached is a request that came back as a
rate-limit error and ended the turn. On an API key you see
API Error: Request rejected (429) instead: that is the key's own per-minute
limit, not a plan window.
/model gets past an Opus or
Sonnet limit, but not the session or weekly limit, which cover every model.
How do you get past a Claude Code usage limit?
| Way out | What it costs |
|---|---|
| Wait for the reset | Time: up to five hours, or up to a week for the weekly limit. |
Usage credits, with /usage-credits | Standard API rates on top of the plan, up to a monthly cap you set. On Team and Enterprise the command asks an admin. |
| A bigger plan | A higher fee for higher limits. The weekly limit still applies. |
| An API key | Every token at API rates, with no window: the account is charged for what it uses, up to any spend limit set in the Claude Console. |
Anthropic's help describes falling back to an API key as temporary, and only where your
organization allows it. Check /status if a key is set in
ANTHROPIC_API_KEY: it can quietly send a subscription session through the key.
How does Selan handle Claude Code rate limits?
Selan runs Claude Code on the Claude accounts and API keys your company already pays for, and reads each Claude account's 5-hour and 7-day windows from every response.
Its statusline has a 5h and a 7d meter for how much of each
window is used. A meter turns amber at 70% and orange at 85%. From 85% it also shows when
its window resets, in your own time zone. A window that is not reported draws no meter,
never 0%. selan usage prints the same windows and reset times without
launching Claude Code.
Once a Claude account's window is spent, Selan does not use that account again until the window resets. Requests go to subscription accounts first and fall through to a key billed per token only when they are capped, so nobody changes a setting on their laptop.
The fallback needs an owner to have connected a key, such as an Anthropic API key, Vertex or OpenRouter, and switched the model on there. A spend limit per person, by day, week or month in UTC, caps what each developer spends on those keys. The fallback does not lift it.
The meters are in the statusline docs, and
selan usage in the reference.
What breaks at team scale?
/usage describes one account, so nobody sees which developer is about to hit
a window until a long job stops. On Team and Enterprise, usage credits go through an
admin, so each request waits on a person. And an API key as the way out means a key on
every laptop, while a Console spend limit caps an organization or a workspace, not one
developer.
In Selan, an owner sees every connected Claude account's 5-hour and 7-day meters on one page, and a spent account stays marked until its window rolls over.
A Selan spend limit stops spending that is billed per token. A subscription is not billed that way, so its ceiling is still Anthropic's window. The limit binds the moment it is reached, and the refusal names the limit and when it lifts.
How a limit per person is set is on the spend page, and keeping keys off laptops in Claude Code without API keys.
Free for up to three seats, and the Free plan needs no card.
The short version
- Two windows: a session limit that resets every five hours, and a weekly limit.
- The message says which limit you hit and when it resets.
- Waiting costs time. Usage credits, an API key and a bigger plan cost money.
What Claude's plans, windows and messages do is as Anthropic's Help Center (usage and length limits, the Max plan, usage credits, models and limits in Claude Code) and its Claude Code docs (the error reference, interactive mode, the statusline and the hooks reference) described them on 13 October 2026. What Selan does is the CLI and the control app as they stood on the same day.