Claude Code limits, explained

Fable 5.2 launched November 9, 2026. How to use it in Claude Code, how it counts against each plan's limits, and the launch week offer that ends November 16.

The session limit resets five hours after you start; the weekly limit resets at a fixed time for your account. They count at once, so you can hit either first.

Yes. claude.ai, Claude Desktop, mobile, Claude Code and IDE extensions all draw from one plan allowance. What that means, and where the exceptions and blind spots are.

On Max and Ultra you can spend up to 50% of your weekly limit on Fable models. It is a share of the same limit, not extra room. What happens at the cap.

A limit reset refills your five-hour, weekly or Fable usage on the spot. How to redeem one, why the button is missing in Claude Code, and when it expires.

The Claude session limit resets in full five hours after your first message. How to see the exact countdown, what happens at the limit, and why verified Ultra has none.

Your Claude weekly limit resets at a fixed day and time assigned to your account, not 7 days after first use and not on your billing date. Here is how to find yours.

Every official place to check your Claude usage: Settings > Usage on web, Desktop and mobile, the Desktop usage ring, Claude Code's /usage and the VS Code dialog.

Anthropic documents why Claude Code usage climbs: resent context, cache misses, subagents, scheduled tasks, compaction, model and effort. Plus what users report.

Yes, but on usage credits, not inside Pro limits. The July 19, 2026 end of the promotion, the Fable 5.2 launch week offer, and the consent prompt in Claude Code.

Max, Ultra and Team include monthly Claude Platform credits, but they do not pay for interactive Claude Code. Usage credits do. Here is what each covers.

Max is $100 and Ultra is $200 as of November 2026. Compare 5x per window against 10x weekly, the five-hour limit, API credits, wrap-up and the Fable share.

Pro is $20 a month and Max is $100 with 5x Pro usage. What else differs for Claude Code: Fable, wrap-up allowance, API credits, and why there are no message counts.

How usage credits let Pro, Max and Ultra keep working past a limit: setup, spend limits, API-rate billing, the Now using credits switch, expiry and the old name.

Anthropic publishes no current hour or message figures for Claude Code on Pro. Why the 2025 estimates are outdated, what changes your usage, and how to measure your own.

Standard seats give 1.25x Pro per-session usage, Premium 6.25x. Prices, Fable access, per-member limits and how to ask an admin for usage credits, as of November 2026.

Verified Ultra accounts have no five-hour limit. Until verified, you get a temporary one. How verification works, why phone checks do not count, and how to check.

Every published multiplier, Max 5x, Ultra 10x weekly, Team 1.25x and 6.25x, is relative to Pro. Here is what each one covers and why none is a message count.

Anthropic renamed Max 20x to Ultra on October 30, 2026 and Max 5x to Max. Same price, automatic move, and new wording of 10x weekly. What changed and what did not.

Your Claude plan renews on a fixed billing date you cannot change. How to cancel in time, app store differences, monthly-only Max and Ultra, and what Headroom shows.

Pro, Max, Ultra, Team and Enterprise all include Claude Code as of November 2026. Free does not. Plus the Console pay-as-you-go route and the April pricing-page test.

CLAUDE_CONFIG_DIR points Claude Code at a different configuration folder. Here is what lives in it, how the macOS Keychain follows it, and where it is ignored.

Give each Claude Code account its own configuration folder, add a shell alias per account, and sign in once per folder. Here are the steps and the traps.

What Anthropic's own documents say about holding several Claude accounts: work and personal setups, sharing, creating accounts to avoid guardrails, and limits.

How to keep an employer's Team seat and your own Claude plan separate in Claude Code and on claude.ai, and what the wrong login method error means.

Sessions in the same folder share one login, but sessions in separate CLAUDE_CONFIG_DIR folders run side by side. Here is how, and what to watch for.

Claude Code has no account switcher. Give each account its own folder and start a different command instead of running /logout and /login each time.

/clear costs nothing and starts fresh. /compact summarizes a long chat but reads all of it first. Here is Anthropic's rule of thumb and when each one is right.

/status shows your version, model, account and connectivity. /usage shows plan limits and costs. Here is which command to run for which question, and why docs disagree.

The habits Anthropic documents to stretch Claude Code usage: clear between tasks, match the model, point at files, lean CLAUDE.md, plan first, lower effort, trim MCP.

Yes. Claude Code in VS Code, Cursor and JetBrains draws from the same plan limits as the terminal and claude.ai. Here is where to see usage and one key trap.

Yes if you are signed in with a Claude plan and no API key is set. If an API key is present, claude -p always uses it and bills the API. Here are the details.

Type /usage in Claude Code to see plan limits, a breakdown of what used them, and a usage-credits row. Here is what each part shows and what the dollar figure means.

Use /model inside a session, claude --model at launch, or ANTHROPIC_MODEL in your environment. Here are the aliases, the default, and the cost of switching.

Run /status in Claude Code to see the account, or claude auth status in a terminal. The config folder a session starts with decides the login. Here is how.

Claude Code can wait in an open session and carry on after a usage limit resets. Here is when it starts, its caps, how to cancel it, and how to turn it off.

Claude Code passes five_hour and seven_day usage to your status line script. Here are the exact fields, who gets them, when they appear, and what is missing.

Haiku uses the least, then Sonnet, then Opus, and Fable uses limits fastest, according to Anthropic. Here is what that means for Claude Code and what is not published.

This Claude Code message is a short server-side throttle, not your plan limit. What it means, how retries work, and how it differs from 429 and 529 errors.

How Team plan limits work per member, what the spend-limit messages in Claude Code mean, how to ask your admin for usage credits, and when a Premium seat helps.

On a Claude plan, "Credit balance is too low" in Claude Code means requests are going to a Console account, often through an API key. How to check and fix it.

What "Login expired · Please run /login" means in Claude Code, the three-day warning before it, how to renew the login, and why long unattended runs stop.

What Anthropic documents about limit warnings and stale bars, what users report about stops near 95% and rounding, and how to check your current figures.

When Claude Code says a single model's limit is used up, a model outside that family keeps you working. Here is when that works, when it does not, and the catch.

What "You've hit your session limit" means in Claude Code, and the documented ways forward: wait, let it continue by itself, add usage credits or upgrade.

The weekly limit message in Claude Code blocks every model until the reset shown. Why automatic continue rarely kicks in, and what you can really do.

If ANTHROPIC_API_KEY is set, Claude Code can use it instead of your subscription and charge API rates. How the precedence works, how to spot it and how to unset it.

Add another Claude Code account to Headroom from its menu: name it, sign in with Claude in the browser, and track it. The Terminal route works too.

A Sign-in lapsed row means that account's Claude Code login can no longer be used. Click Sign In Again… in Headroom, or run /login in that account's folder.

What Ready, N% of session left, Back in 38m and a model's back in status mean in Headroom, and how each maps to Claude's session, weekly and per-model limits.

Headroom is a free Mac menu bar app that shows session and weekly use for each Claude Code account, with reset countdowns, plan and an estimated renewal date.