Claude Code usage limits explained

"Claude Code usage limit" means a different thing depending on how you sign in. A Pro subscriber, a Team seat and a Console API key each have a different ceiling, and the message you see names which one you reached. This page covers each, with every figure taken from Anthropic's docs as read on 2026-10-02.

Which limit applies to youCopy Anchor Link

Run /status to see which credential a session is using. If an ANTHROPIC_API_KEY is set in your environment, Claude Code uses it instead of your subscription, so you get API charges and API rate limits rather than your plan's allowance (Anthropic support).

Pro and Max: the 5-hour and weekly windowsCopy Anchor Link

Pro and Max plans include a rolling usage allowance shared across Claude and Claude Code: chat, the terminal and the IDE extensions all draw on the same limits. Usage counts against two windows at once, a 5-hour session window and a weekly window (error reference). When one runs out you see one of:

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
You've hit your Sonnet limit · resets 3:45pm

• Session and weekly limits cover every model, so switching with /model does not help. One heavy burst, such as a large fan-out of subagents, can use up the weekly window before the session window resets.
• Opus and Sonnet limits apply only to that model family. Switching to a model outside it keeps you working, though the first request after the switch re-reads the whole conversation with no cache hits, because each model has its own cache.

What you can do:

• See where you stand. /usage shows your plan bars and reset times, and a breakdown of what drove your usage: skills, subagents, plugins, MCP servers and flags such as long context or cache misses. The breakdown covers this machine only (Manage costs). The status line can show both windows continuously through its rate_limits.five_hour and rate_limits.seven_day fields (status line docs).
• Wait it out. Recent versions of Claude Code can wait in the open session and continue the interrupted task shortly after the reset.
• Turn on usage credits with /usage-credits to keep working past the allowance, billed separately, optionally up to a monthly spend limit you set.
• Upgrade from Pro to Max 5x, or Max 5x to Max 20x, if you hit the limit regularly. Plan prices are on claude.com/pricing.

Team and Enterprise seatsCopy Anchor Link

On Claude for Teams and Enterprise, each member's Claude Code usage draws on a per-seat allowance with the same 5-hour and weekly windows, shared with Claude chat and Cowork. Its size depends on the seat tier, Standard or Premium (Manage costs).

Past the allowance, a member can keep working only if the organization has turned on usage credits, and then only up to the spend limits an admin set for the organization, a group or that member. Those produce different messages:

You've hit your individual spend limit · ask your admin for a higher limit
You've hit your org's monthly spend limit · visit claude.ai/admin-settings/usage to raise it
You've hit your team's shared budget · ask your admin to raise it at claude.ai/admin-settings/usage

/usage-credits sends the member's request for more to the admins. An admin raises the limit under Admin settings > Usage on claude.ai. When the message also names a plan reset time, waiting for it works too. Setting those limits is covered in how to budget Claude Code for a team.

API keys: rate limits and monthly spend capsCopy Anchor Link

With a Claude Console API key there is no 5-hour window. Two other limits apply (rate limits):

• Rate limits in requests per minute, input tokens per minute and output tokens per minute, per model class. Going over returns a 429 with a retry-after header, and Claude Code retries automatically. The limits use a token bucket, so capacity refills continuously rather than at a fixed time. For most current models, cache reads do not count toward the input tokens limit, which matters for Claude Code because most of a long session's input is cache reads (see cache tokens explained).
• A monthly spend cap set by the organization's usage tier: $500 on Start, $1,000 on Build and $200,000 on Scale, with no cap on Custom. At the cap, requests stop until the next calendar month. You can set your own lower limit on the Console Billing page, and per-workspace spend and rate limits too.

The first time you sign in to Claude Code with a Console account, a workspace called "Claude Code" is created for its usage. On an organization with custom rate limits, Claude Code traffic counts toward the organization-wide limits, so a workspace rate limit on that workspace stops it crowding out production traffic (Manage costs).

A Credit balance is too low error means the Console organization has run out of prepaid credit, or that Claude Code is using an API key you did not mean it to use.

Messages that are not your limitCopy Anchor Link

• Server is temporarily limiting requests (not your usage limit) is a short server-side throttle. Claude Code retries it automatically; if it persists, check status.claude.com.
• A context or auto-compact warning means the conversation is close to the context window, not that your allowance is spent.
• Usage credits required for 1M context is an entitlement check for the 1M-token context window. It fires even when your allowance has room left. Turn on usage credits with /usage-credits and restart, or pick the model without the [1m] suffix in /model.

Why your limit runs out faster than you expectCopy Anchor Link

Every request re-sends the whole conversation. A one-line question in a session that has been open all day still draws usage for everything above it, and Anthropic lists the usual reasons a long session climbs (Manage costs):

• Long context. Run /clear between unrelated tasks. It costs nothing, where /compact is itself a large request.
• Cache misses. Coming back after more than the cache lifetime (an hour on a subscription, five minutes on usage credits or an API key by default) reprocesses the full context.
• Background work. Scheduled /loop tasks, subagents, agent teammates and goal check-ins each send requests on top of yours.
• Model and effort. Opus at high effort spends more of the allowance per request than Sonnet at a lower one; thinking tokens are output tokens.

Seeing what used your allowanceCopy Anchor Link

Plan limits are shown as a percentage, and the /usage breakdown of what used them only sees the machine it runs on. If you work across a laptop, a dev box and cloud sessions, nothing adds them up. SessClone's Collector reports each Turn's tokens and model from every machine to one place, so you can see which Sessions, Projects and models the week went on, and what the same work would have cost on an API key. It is an estimate, not your plan's meter.

Install the Collector to start, or read Claude Code cost per session for how a Turn's tokens become dollars.