Claude Code usage limits, plan by plan

One pool with claude.ai

Claude Code is included in every paid plan and "shares the same usage limits as the rest of your plan, so your work in the terminal and your chats draw from one pool". The terminal, the IDE extensions (VS Code and its forks, JetBrains), Claude Desktop's Code tab and the web all count toward the same session and weekly limits as your chats. Free has no Claude Code.

Anthropic sizes some of it separately for Claude Code, though:

• 6 May 2026: Claude Code's five-hour limits doubled on Pro, Max, Team and seat-based Enterprise, and the peak-hours reduction stopped applying to Claude Code on Pro and Max.
• 13 May to 13 September 2026: a promotion raised Claude Code's weekly limits by 50%.
• 14 September 2026: Claude Code's weekly limits settled 25% above their level before the promotion, about 17% below the promotion, as Anthropic itself pointed out.
• 22 September 2026: with Claude Opus 5.5, all five-hour limits of Pro, Max, Team and seat-based Enterprise went up; Anthropic's Claude Code account put the increase at 20% for Claude Code's five-hour session limits.

How Claude's limits changed

7 changes between March 2026 and September 2026. Dates in UTC.

1. 1Mar 13 → Mar 28Off-peak promotion: five-hour usage doubled outside weekday peak hours
2. 2Mar 26Peak hours: five-hour limits drain faster on weekdays, 5–11 AM PT
3. 3May 6Claude Code: five-hour limits doubled, no peak-hour reduction on Pro and Max
4. 4May 13 → Sep 13Promotion: Claude Code weekly limits 50% higher
5. 5Jul 20Fable 5: up to 50% of weekly limits on Max, usage credits only on Pro
6. 6Sep 14Claude Code weekly limits permanently 25% above their pre-promotion level
7. 7Sep 22Opus 5.5 launch: five-hour limits raised on Pro, Max, Team and Enterprise seats · One limit reset given to subscribers,…

• Promotion
• Less usage
• More usage
• Model rules

The limits that can stop Claude Code

Claude Code names the limit that stopped it and when it comes back:

• "You've hit your session limit · resets 3:45pm": the five-hour session, shared with claude.ai.
• "You've hit your weekly limit · resets Mon 12:00am": the week, all models.
• "You've hit your Opus limit" or "You've hit your Sonnet limit": that model family only. The session and weekly limits are shared by all models, so switching model does not help there; for an Opus or Sonnet limit, /model to another family keeps you working.

Before a window runs out, Claude Code can warn you: "You've used 85% of your session limit · resets 3:45pm". Usage counts against the session and the week at the same time, and Anthropic notes that "a single burst of heavy activity, such as a large workflow fanout, can exhaust the weekly allowance before the session window resets".

The session, the week and a model limit: any of them can stop a request

Example: session 72%, week 64%, Opus 100%. Whichever fills first stops you.

Example

Blockedonly Opus is blocked: switch model and keep working.

Limits in this example

Limit	Used	Resets
5-hour session	72%	—
Weekly · all models	64%	—
Weekly · Opus only	100%	—

Fable in Claude Code

On Max and premium Team seats, Fable 5 and Fable 5.1 can use up to 50% of the weekly limits, then usage credits or another model. On Pro and standard Team seats, Fable runs on usage credits from the first request. Fable 5 needs Claude Code 2.1.170 or later, Fable 5.1 needs 2.1.255 or later.

Where Claude Code shows your usage

/usage shows your plan's limits, then what used them on this machine

Example output. The bars are your account's limits; the breakdown under them only counts this machine's history.

Illustrative

What each command shows

Command	Shows
/usage	your plan's limits and what used them

• /usage (with /cost and /stats as aliases): your plan's usage bars with reset times, then a breakdown by skill, subagent, plugin and MCP server. Anthropic warns that those figures "are approximate and computed from local session history on this machine, so usage from other devices or claude.ai is not included".
• The status line: rate_limits.five_hour and rate_limits.seven_day give the percentage used and the reset time, for Pro and Max subscribers, after the first response of the session.
• Claude Desktop: the usage ring next to the model picker.

The Session block of /usage shows a dollar figure meant for API users; on Pro and Max, Anthropic says it "isn't relevant for billing purposes".

When a limit stops Claude Code

• Wait, and let it continue: since version 2.1.234, a session signed in with a claude.ai subscription can wait for the reset and pick the task up again, showing "Usage limit reached · continuing automatically at 3:45pm · esc to cancel". Turn it off in /config.
• Usage credits: on Pro and Max, /usage-credits opens Settings > Usage to turn them on and set a monthly spend limit; usage past the plan is billed at API rates.
• Another model, when only an Opus or Sonnet limit is full.
• A limit reset, when Anthropic has given you one: it works from the web or Claude Desktop, not from the terminal, and the refilled limits apply to Claude Code too.