What happens when your team's Claude credits run out

1. Read the message. "You've hit your session limit" or "You've hit your weekly limit" is your own seat's allowance and resets by itself. "You've hit your individual spend limit", "You've hit your org's monthly spend limit" or "You've hit your team's shared budget" is a cap your company set on paid usage, and only an admin can raise it before the next month (Claude Code's error reference).
2. If the message also says when your session limit resets, your included usage comes back then without anyone raising anything. On a subscription, Claude Code waits and continues the task by itself after a 5-hour reset unless you choose otherwise in /rate-limit-options.
3. If you cannot wait, run /usage-credits in Claude Code. For a member without billing access it asks you to confirm and then sends a request to your organisation's admins; running it again does not send a duplicate. On seat-based Enterprise plans the Claude app also shows a "Request usage credits" link.
4. Tell your lead what you were working on. If the whole organisation's limit ran out, other people are stopped too, and the admin needs to know whose work is waiting.

The guide on how to ask your company for more Claude Code budget has a short template for the request.

Usage credits do nothing until an Owner or Primary Owner enables them, for the whole organisation or for chosen members, according to Anthropic's Team usage-credits article. With them off there is nothing to run out except each seat's own window: members "will need to wait for them to reset", and the team's bill does not move. With them on, the organisation's limit is the shared pot. On a Team plan, Owners buy the credits in advance ("Owners can pre-purchase usage credits that they can control using spend limits"), so the balance itself can also run low; the article does not say what members see when it reaches zero. On a self-serve Enterprise plan billed by usage, Anthropic says "When the balance hits zero, usage stops for the whole organization until an Owner purchases more."

Anthropic's Team article is not consistent about when a member who hits their own spend limit can work again: one section says they wait for their usage limits to reset, and its FAQ says they wait until the next billing period or until the limit is raised. Claude Code's error reference sides with the first: when a seat window is what ran out, access returns when it resets. The plain reading is that included usage comes back on the seat's schedule, and paid usage past it waits for the month or for an admin.

Of the 138 people we found on Reddit who said they ran out on an account their employer pays for, 16 described shared team credits that ran out before the month did, usually because a few people used most of them. In their words, without names:

• "We get $500 claude credits per month and we exhaust it in 15 days and it is painful process to get more credits."
• "I got token limit reached for team, now we have to request head of our department for more quota."
• "Is there a way that the admin of our Claude subscription can move tokens from the inactive users to me since I need more?"
• On GitHub Copilot, after a switch to one shared pot: "At the beginning of this month, the company decided to switch us from per-developer licenses to a shared, per-organization quota. Exactly two and a half weeks later the quota has been exceeded."

Across all 138, what people say they do next depends on who is speaking. Engineers mostly ask for more, ration what is left, or stop using the tool until the reset. Managers more often switch vendor or raise the cap. Finance and IT staff set or raise caps, and say they cannot tell whose request is justified.

• Raise the organisation's limit for this month, in Organization settings > Usage, or set it to unlimited. Everyone who was stopped can carry on at once.
• Give individual limits instead of relying on one shared pot. Owners can set a monthly spend limit for each member, so one person's heavy month does not use up everyone's credits. On Enterprise, limits can also be set per group.
• Answer the requests. Members can send a request from Claude Code with /usage-credits, and on seat-based Enterprise from a link in the app; admins review them in Organization settings > Usage and get a daily email of outstanding requests.
• Move the people who run out every day to a Premium seat. On claude.com/pricing a Premium seat has five times the usage of a Standard seat, at a fixed price, which can cost less than the same person's usage credits at API rates.
• Turn usage credits off. The seat windows then become hard stops and the bill stops moving; the cost is the time people spend waiting.

To choose between these you need to know who used the budget and on what. The members list in Organization settings > Usage has each member's month-to-date spend, and the spend report lists spend by user and model, refreshed daily with a one-day delay. Neither shows each member's 5-hour or weekly window, or what the sessions that used the credits were doing.

OpenReason Intel is a dashboard for Claude Code and Codex usage. One install command on each engineer's machine reports their 5-hour and weekly windows and their usage past the plan. In the dashboard a lead can see each person's windows over time, each period past the limit with the session, repository and model around it, the turns underneath with their estimated share, and seat commitments beside variable spending. An engineer who installs it alone sees the same for themselves, which is the record to bring when asking for more.

It does not stop anyone, notify anyone, or add usage. It is a record to look at when deciding what to change. Costs past the plan are shown as reported by the vendor, allocated from vendor amounts, or estimated at list price, and each figure says which.