Why Do I Hit My Claude Code Usage Limit So Fast?
Because Claude Code sends your entire conversation with every request, a session that has been open for hours uses far more than the questions you typed. Anthropic's docs also say your Claude chats and Claude Code draw from one shared pool, so everything you do in both counts against the same limits.
Most of the fixes cost nothing. This post sticks to what Anthropic's help center and docs say, in the order I'd apply it. Limits can change, so check the help center if a number or a command looks different.
How the Limits Work
Paid plans have two kinds of usage limit. A session limit resets every five hours, and weekly limits reset once a week. The five-hour period starts when you send your first message, and the limit resets in full five hours later. The weekly limit resets at a fixed time each week that is assigned to your account, no matter when you start or when your subscription began.
Max and Ultra plans also have a separate weekly limit for Fable, the top model. Verified Ultra accounts have no five-hour limit at all. There is no fixed message count on any plan. Anthropic says what you can do depends on the length and complexity of your conversations, the features you use, the model you choose, and the effort level you select.
To see where you stand, open Settings > Usage on claude.ai. Pro, Max, Ultra and Team plans show progress bars for your five-hour session and your weekly limits there, along with the time remaining before each resets.
What Actually Uses Up Your Usage
Every turn in Claude Code sends three things to the model: the conversation so far, your project context (your CLAUDE.md file plus any files Claude has read), and your new message. Anthropic's help article says the first item grows fastest, and that a long debugging session in which Claude has read twenty files and produced fifteen changes carries all of that on every later message.
Here are the things Anthropic's docs name as the biggest drains:
- A long conversation. Claude Code re-reads that history at the cached rate, which is cheaper than new content, but a one-line question in a session that has been open all day still draws usage for the whole conversation.
- A cold start after a break. Your first message after a break longer than the cache lifetime misses the cache and reprocesses your full context.
- The model. The help center says Opus uses meaningfully more of your quota than Sonnet, and costs several times more per turn, and Sonnet costs more than Haiku.
- Effort and thinking. Higher effort uses more tokens, which are the small pieces of text Claude reads and writes. Thinking tokens count as output.
- Subagents. A subagent is a helper copy of Claude that takes a side task, and every subagent sends its own requests on top of your main conversation's. Agent teams use about 7x more tokens than standard sessions.
- Tools you've connected. Anthropic says tools and connectors are token-intensive.
- Background work. A scheduled task fires on its interval even while your session sits idle, and session recap and idle compaction can use tokens while you're away. A compaction, which is Claude summarizing your conversation, is itself a large request.
How to Stretch Your Usage
Apply these in order. The first three do most of the work.
- Clear between tasks. Type /clear in Claude Code and press Enter whenever you switch to unrelated work. Anthropic calls it the single most effective lever for both quality and cost, and it costs nothing. Your CLAUDE.md and project files stay, and only the chat history goes. It can't be undone, so run /compact instead if you might need something from the history, or name the session with /rename first and come back to it with /resume.
- Use /compact when you're mid-task. It summarizes the conversation into a short recap so you can keep going. Be aware that it reads the whole conversation to summarize it, so compacting a huge conversation is itself a big request. You can tell it what to keep, like /compact focus on code samples and API usage.
- Match the model to the job. Anthropic's defaults: Sonnet for most work, Opus when you're genuinely stuck or the change is wide, and Haiku for quick mechanical tasks. Type /model to switch. A built-in option, /model opusplan, uses Opus while Claude plans and Sonnet while it executes.
- Lower the effort. Type /effort to dial it down when a task is simple.
- Ask for a plan first. For anything touching more than two or three files, press Shift + Tab to cycle to plan mode, or ask Claude to list the files it'll touch before it changes anything. A plan costs a few hundred tokens, and a wrong change you revert and redo costs far more.
- Be specific. 'Improve this codebase' triggers broad scanning. 'Add input validation to the login function in auth.ts' lets Claude read only what it needs.
- Point at files instead of pasting them. Anything you paste sits in context for the rest of the session. Anthropic notes that typing @ before a file name pulls in the entire file plus its CLAUDE.md tree, so use the bare file path when your goal is to save usage. Trim logs to the 20 or 30 relevant lines first.
- Keep CLAUDE.md short. It's sent with every turn. Anthropic suggests staying under roughly 200 lines, and moving instructions for specific workflows into skills, which only load when you use them.
- Turn off connections you aren't using. Type /mcp to see your connected tools and disable the ones you aren't using, and type /context to see what is taking up space.
- Catch mistakes early. Press Escape the moment Claude heads the wrong way, and use /rewind (or double-tap Escape) to go back to an earlier checkpoint. Give Claude a way to check its own work, like a test or the result you expect.
- Delegate noisy jobs. Running tests or reading long logs inside a subagent keeps the bulky output out of your main conversation, but remember the subagent's own requests still draw on your usage. You can pick a smaller model for it. Keep agent teams small, and shut teammates down when their work is finished.
How to See Where Your Usage Went
Type /usage in Claude Code. For Pro, Max, Ultra, Team and Enterprise plans, Anthropic's docs say its breakdown recommends a change for behaviors behind 10% or more of your recent usage, like long context and cache misses. The same screen shows your plan usage bars.
/cost is the same screen as /usage; the session cost figure there isn't relevant to billing on Pro, Max and Ultra.
What to Do When You Hit the Limit
- Wait. The limit resets in full when your five-hour window ends, and your weekly limit resets at its fixed time.
- Watch for the wrap-up. If you hit the five-hour limit while Claude Code is partway through a response, it may keep working briefly so it can finish the step in progress, and you'll see Usage limit reached, wrapping up. Anthropic says this allowance is small and capped, the usage counts toward your weekly limit, and it's rolling out gradually, so you may not have it yet.
- Turn on usage credits. On paid plans, usage credits let you keep working at standard API rates. Type /usage-credits in Claude Code to open the setting. The monthly API credits that come with Max and Ultra can't be used for this, or in Claude Code at all.
- Move up a plan. Anthropic suggests Max if you consistently hit limits on Pro, and Ultra if you consistently hit them on Max. Your new plan's limits apply right away.
- Use a Claude Console account. This is pay-as-you-go, billed at standard API rates.
Common Problems
- You hit the limit after barely using Claude Code today: your Claude chats on the web, desktop and mobile share the same pool, so check those too.
- You're being charged per use instead of using your plan: a saved Anthropic API key makes Claude Code bill that key instead of your plan. Anthropic has a help article called Managing API key environment variables in Claude Code. If you're signed in with a Console account, type /login in Claude Code to switch to your subscription.
- Usage seems to drain while you're away: scheduled tasks, session recap and idle compaction can all run while a session sits open. Close sessions you aren't using.
- Claude says the context window is full: that's a different limit from your usage limit. Type /compact to keep going, or /clear if you don't need the old history.
Where to Go Next
Check /usage after a few sessions to see which habit is costing you most, fix that one first, and only think about a bigger plan once your habits are tight. Plan mode, a short CLAUDE.md and /clear between tasks are the three habits worth building first. If you'd like them built into your first projects, Claude Code Club is $9/month, and that's where I walk members through them.
Short, practical drops on skills, MCP, agents, prompts, and more. No spam, unsubscribe anytime.
Frequently asked questions
How long until my Claude Code usage limit resets?
The five-hour limit starts when you send your first message and resets in full five hours later. Weekly limits reset at a fixed time each week assigned to your account. You can see your next reset time in Settings > Usage on claude.ai.
Do Claude chat and Claude Code share the same usage limit?
Yes. Anthropic's help center says usage across claude.ai, Claude Code and Claude Desktop counts toward the same limit, so a long chat in the Claude app leaves less for Claude Code.
What happens when Claude Code hits the usage limit?
Claude Code stops taking new work until your window resets, and your finished work is not lost. You may see a short wrap-up allowance first so it can finish the step in progress.
How many messages do I get on Pro or Max?
There's no fixed message count. Anthropic says it depends on the length and complexity of your conversations, the model you choose, the features you use and the effort level. Max gives 5x the usage of Pro every 5 hours and every week.
Can I get unlimited usage?
Anthropic's answer is usage credits on paid plans, which keep you working at standard API rates after you reach your limit. You can also use a Claude Console account, which is pay-as-you-go. Both cost money beyond your plan.
Last reviewed by Duncan Rogoff on November 9, 2026
