Claude Code “Usage Limit Reached”: Reset Times and Fixes
Claude usage limit reached. Your limit will reset at …
For companies: one company account, with a key and a fixed monthly budget for every person and app. Create a company workspace → or talk to us
Claude Code stops with a usage-limit message when your Pro or Max plan's 5-hour session window or weekly limit is spent, shared with Claude chat, and the message tells you when it resets; run /usage to see both windows. On Ultra (the $200 plan called Max 20x until October 30, 2026) only the weekly limit applies once your account is verified. To keep working you can wait for the reset, redeem a saved limit reset if you have one, buy usage credits at API rates, switch to a pay-per-token Anthropic API key, or point Claude Code at a flat-rate Anthropic-compatible API such as Standard Compute (ANTHROPIC_BASE_URL set to https://api.stdcmpt.com plus the model variables, from $19/month), which has no 5-hour or weekly windows and a fixed monthly compute budget instead.
What causes it
- The 5-hour rolling window: all Claude usage (chat + Claude Code) shares it.
- Weekly caps on top of the session window for sustained heavy use.
- Large repositories and long sessions: context grows, so each step consumes more of the budget.
- “You've hit your monthly spend limit · raise it at claude.ai/settings/usage” is a different stop: a window is spent and your usage credits reached the monthly cap you set. Raise it under Settings → Usage → Usage credits, or run /usage-credits.
How to fix it
- Wait for the window reset (the error tells you when), or redeem a saved limit reset in Settings → Usage → Resets if Anthropic gave you one.
- Use /compact regularly and start fresh sessions per task to keep context small.
- Upgrade the plan tier (Max raises both caps; verified Ultra accounts have no 5-hour limit, only a weekly one) or switch Claude Code to API billing for pay-as-you-go.
- Point Claude Code at a different Anthropic-compatible endpoint: set ANTHROPIC_BASE_URL, ANTHROPIC_AUTH_TOKEN, and ANTHROPIC_MODEL plus the ANTHROPIC_DEFAULT_*_MODEL slots to model IDs that endpoint serves, and your claude.ai plan limits no longer apply to those requests.
Running an agent? Agentic loops are what burn the window: every tool result re-enters context. The same task done with smaller context lasts several times longer before hitting the cap.
Sources: Claude Code docs: errors; Claude Help: usage credits; Claude Help: what is the Ultra plan?; Claude Code docs: LLM gateway
Compare APIs for Claude Code
Best API for Claude Code →
4 options compared, with prices and limits.
Keep working on a fixed monthly plan
Standard Compute serves the Anthropic Messages API (/v1/messages), so Claude Code keeps working on it: run standardcompute launch claude, or set ANTHROPIC_BASE_URL to https://api.stdcmpt.com (no /v1), ANTHROPIC_AUTH_TOKEN to your key, and ANTHROPIC_MODEL plus ANTHROPIC_DEFAULT_OPUS_MODEL, ANTHROPIC_DEFAULT_SONNET_MODEL and ANTHROPIC_DEFAULT_HAIKU_MODEL to catalog IDs such as anthropic/claude-sonnet-5.5 (Claude Code's own claude-sonnet-5-5 style names return model_not_found, so leave no slot empty). Plans are a fixed monthly price from $19 with a fixed compute budget and no 5-hour or weekly windows; requests stop with a 402 when the month's budget is used, with no overage charges. Claude Opus 5.5, Sonnet 5.5 and Fable 5.2 are in the catalog next to GPT, DeepSeek, GLM and Kimi; smart routing picks per request unless you pin a model. It does not include a Claude Pro or Max subscription.
standardcompute launch claude
# Or set them yourself (no /v1 here):
export ANTHROPIC_BASE_URL=https://api.stdcmpt.com
export ANTHROPIC_AUTH_TOKEN=YOUR_API_KEY
export ANTHROPIC_API_KEY=
export ANTHROPIC_MODEL=anthropic/claude-sonnet-5.5
export ANTHROPIC_DEFAULT_OPUS_MODEL=anthropic/claude-opus-5.5
export ANTHROPIC_DEFAULT_SONNET_MODEL=anthropic/claude-sonnet-5.5
export ANTHROPIC_DEFAULT_HAIKU_MODEL=anthropic/claude-haiku-5.5
claudePlans from $19 a month with $19 of compute. Eligible new accounts can try $0.25 of compute first, no card. Trial terms
FAQ
How long until Claude Code unblocks?
The session limit resets every five hours, and the message shows the exact reset time. The weekly limit resets at a fixed day and time assigned to your account, the same every week, shown in Settings > Usage.
I'm on Claude Ultra. Why does Claude Code still hit a 5-hour limit?
Because your account is not verified yet. Ultra has no five-hour limit for verified accounts, but a new or unverified one gets a temporary limit until Anthropic verifies it automatically from account and payment history. Check Settings > Usage: a Current session bar means the temporary limit still applies. Once it is gone, only the weekly limit (10x Pro's usage, plus a separate Fable limit) stops Claude Code.
What do Claude Max users switch to when they keep hitting the weekly limit?
Three things: usage credits on top of Max, billed at API rates; a Claude Console API key, billed per token with no weekly limit; or a flat-rate Anthropic-compatible API such as Standard Compute, which keeps Claude Code and replaces the session and weekly windows with one monthly compute budget. The API key never cuts you off but has no fixed price; the flat-rate plan has a fixed price but stops when its budget is used.
Can I point Claude Code at a different provider to avoid limits?
Yes. Claude Code speaks the Anthropic Messages API, and any endpoint that serves it works through ANTHROPIC_BASE_URL plus a key in ANTHROPIC_AUTH_TOKEN, with ANTHROPIC_MODEL and the ANTHROPIC_DEFAULT_*_MODEL slots set to model IDs it serves. While that key is set, Claude Code sends it instead of your claude.ai login, so Pro and Max limits don't apply. Anthropic's docs say it doesn't support routing Claude Code to non-Claude models through a gateway, so test a real task first.
Can I use Opus 5.5 or Fable in Claude Code without the weekly cap?
Yes, on any per-token or budget-based route: a Claude Console API key, Amazon Bedrock, Google Cloud or Microsoft Foundry bill per token, and Standard Compute serves anthropic/claude-opus-5.5 and anthropic/claude-fable-5.1 against a monthly budget. Opus 5.5 costs $4/$20 per million input/output tokens and Fable 5.1 $10/$50, so pinning them all day uses a budget fast.
Related errors
For companies: one company account, with a key and a fixed monthly budget for every person and app. Create a company workspace → or talk to us