Claude usage limit reached: what are my options?

The message names the limit, and each one has its own way out (Claude Code docs: errors). Session and weekly limits are shared across all models, so switching models does not restore access. Paid plans have a session limit that resets every five hours and a weekly limit, and usage on claude.ai and the desktop app counts toward the same limits as Claude Code (Help Center: usage and length limits). The Opus and Sonnet limits apply only to that model family.

Max plans also have a separate weekly limit for Fable (Help Center: Claude Code with Pro or Max). Anthropic's pricing page lists Fable on Max as "50% of weekly limits" and on Pro under usage credits. Depending on your plan and seat tier, Fable usage can bill to usage credits instead of your plan's limits, and Claude Code asks before it does. You can then continue on Fable with usage credits or switch to your default model (Claude Code docs: Fable and usage credits).

A 529 Overloaded error is not a usage limit. Anthropic's error reference says it does not count against your quota, and suggests waiting a few minutes or switching models.

There are four ways to keep going. Which ones apply depends on the limit and on how you sign in. Run /usage first to see your limits and when each one resets.

• Wait in the open session. On Claude Code v2.1.234 or later, signed in with a claude.ai subscription, Claude Code waits after a usage limit and continues the task on its own when the limit resets, as long as the session stays open (Claude Code docs: wait for a usage limit to reset). It does not start that wait on its own when the reset is more than 24 hours away, which a weekly limit can be, or for an Opus or Sonnet limit while you run another family. In those cases pick "Wait here, then continue automatically" from /rate-limit-options. The continued task still asks for permissions as usual, so it can stop on a prompt while you are away. To turn the automatic wait off, set autoContinueAtUsageLimit to false or use /config.
• Turn on usage credits. Run /usage-credits while signed in with your claude.ai subscription. On Pro and Max it opens Settings > Usage on claude.ai, where you turn usage credits on, add funds and set a monthly spend limit (Claude Code docs: add usage credits). Usage credits are prepaid, billed at standard API rates separately from the subscription, and cover Claude Code as well as chat (Help Center: usage credits). If you add them with /usage-credits during an automatic wait, Claude Code checks whether usage is available and continues right away if it is. On Team and Enterprise without billing access, the command sends a request to your admins.
• Switch model family, for an Opus or Sonnet limit only. Run /model and pick a model outside the limited family. Each model has its own prompt cache, so the first request after the switch reads the whole conversation again with no cache hits.
• Use an API key. A key from the Claude Console is a separate bill, per token, with no reset to wait for. Its spend shows on the Console usage page, and the /usage-credits command is not available with API key sign-in.

The monthly API credits included with Max plans cannot be used in Claude Code or the Claude apps (Help Center: Claude Code with Pro or Max).

Upgrade if you hit the limit most weeks, and use usage credits if it happens now and then. As of 23 October 2026, Anthropic's pricing page describes Max as "5x or 20x more usage than Pro" and gives the limits as multiples, not token counts.

Anthropic's Help Center suggests moving from Pro to Max 5x, or from Max 5x to Max 20x, if you consistently hit limits (Help Center: Claude Code with Pro or Max). Usage credits bill only what your plan does not cover, such as usage past a limit. A larger plan costs more every month. Compare this month's credit spend with the price difference on the pricing page. Settings > Usage on claude.ai shows that spend, and /usage shows it only once you have set a monthly spend limit (Claude Code docs: check your usage-credits spend). Prices and plans change, so check the page on the day you decide.

Usage credits are billed at standard API rates (Anthropic: API pricing). Once Claude Code draws on them it drops the main conversation to the five-minute cache unless you set promptCacheTtl (Claude Code docs: which TTL each request gets), so the first request after a break longer than five minutes reprocesses the full context. For scale, one 120-prompt session of Claude Code on its own, on an open-source dashboard, cost $104.71 at the Claude Opus 5.5 list prices we used in our runs (one run, our benchmarks). Those prices are $4 input, $20 output, $0.20 cache read and $8 cache write per million tokens. The $8 is the one-hour write rate. A five-minute write is $5 (Anthropic: pricing). Our figures price cache writes at that one-hour rate, as Claude Code's own cost readout totals them.

Find what uses the most, then cut it. Anthropic's cost docs list what makes usage climb, and the first cause they list is that Claude Code sends the full conversation with every request (Claude Code docs: why usage climbs in a long session). Our guide on why Claude Code usage runs out so fast goes through each cause.

• Read /usage on a Pro, Max, Team or Enterprise plan. It attributes recent usage to skills, subagents, plugins and MCP servers, and suggests changes. Press d or w to switch between the last 24 hours and the last 7 days. The figures are approximate and come from this machine's history only.
• Run routine work on Sonnet. The cost docs say Sonnet handles most coding tasks well and costs less than Opus, and suggest keeping Opus for complex architectural decisions. On Max, Fable has its own weekly limit, so keep it for the work that needs it.
• Run /clear between unrelated tasks. It costs nothing, and the next task does not carry the last one's history. /compact reads the conversation it summarizes, so compacting a large context is itself a large request.
• Watch what runs while you are away. Scheduled tasks, messages from your other sessions and goal check-ins each send the full context, and every subagent sends its own requests.
• Mind the cache lifetime. The first message after a break longer than the cache lifetime reprocesses the full context (how to check and set the lifetime).