Claude Code and Codex meter your subscription in time windows — a five-hour window plus a weekly limit — while Cursor uses monthly usage pools that reset with your billing cycle. Claude Code shares one allowance with Claude on web, desktop and mobile; Max 5x and Max 20x give 5× and 20× the Pro per-session allowance. Codex shares one allowance with ChatGPT Work; Plus and Business have five-hour and weekly windows, while Pro tiers currently have no five-hour limit. Cursor Pro, Pro Plus and Ultra include a Cursor Models pool and an Other Models pool, and unused usage doesn't roll over. None of the three publishes a fixed message count you can rely on, so check the official usage screen for your live numbers.
Last updated October 12, 2026 · Sources checked October 12, 2026
Key facts
| Claude Code: meter | Five-hour session limit + weekly limit on Pro and Max; Max also has a separate weekly limit for Fable |
|---|---|
| Claude Code: shared with | Claude on web, desktop and mobile — all surfaces count against the same limits |
| Claude Code: where to check | /usage in Claude Code; next reset time in claude.ai Settings › Usage |
| Cursor: meter | Two monthly pools: Cursor Models (Grok 4.7, 4.6, 4.5, Composer 2.5) and Other Models (third-party models at provider API prices) |
| Cursor: reset | Monthly with your billing cycle; unused usage does not roll over |
| Cursor: where to check | Spending tab in the Cursor dashboard (shows both pools and the reset date) |
| Codex: meter | Five-hour window + weekly window on Plus and Business; Pro 100 / 200 / 500 currently have no five-hour limit |
| Codex: shared with | ChatGPT Work (and ChatGPT for Excel / Workspace Agents where your plan has them) |
| Codex: where to check | Settings › Usage, or /status in the Codex CLI |
| Codex: can Support reset it? | No — OpenAI Support does not reset ChatGPT or Codex usage limits |
How the three systems differ
| Claude Code | Cursor | Codex | |
|---|---|---|---|
| How usage is metered | Five-hour session limit + weekly limit (Max: + weekly Fable limit) | Two monthly pools, drawn down at each model's price | Five-hour window + weekly window (Pro tiers: currently no five-hour limit) |
| When it resets | Session every 5 hours; weekly once a week at a fixed time assigned to your account | With your monthly billing cycle; no rollover | New 5-hour window starts at your first message after the last one ends; weekly period per Settings › Usage |
| Shared with | Claude chat, Claude Desktop, IDE extensions | Cursor only (both pools) | ChatGPT Work |
| Plans with limits | Pro, Max 5x (5× Pro per session), Max 20x (20× Pro per session) | Pro, Pro Plus, Ultra (both pools); Start, India only (Cursor Models only) | Plus, Pro 100 / 200 / 500, Business (Free and Go also include Codex) |
| Where to check | /usage; Settings › Usage | Spending tab | Settings › Usage; /status in the CLI |
| Keep going past the limit | Usage credits at standard API rates, or upgrade | On-demand usage at the same API rates, or upgrade | Credits, a banked reset, a purchased instant reset (eligible Plus/Pro), or upgrade |
| Own API key | An ANTHROPIC_API_KEY env var makes Claude Code bill the API instead of your plan | Individual plans: BYOK requests don't draw from either pool | API usage is billed separately at API rates |
The practical difference: with Claude Code and Codex you can run out for a few hours and come back; with Cursor, once your included monthly pool is spent you stay on pay-as-you-go (if enabled) until the next billing cycle.
Claude Code usage limits (Pro and Max)
Per Anthropic's help center, both Pro and Max have a five-hour session limit and a weekly limit, and Max plans also have a separate weekly limit for Fable. These limits are shared across Claude and Claude Code: chat on claude.ai, Claude Desktop, Claude Code in the terminal and the IDE extensions all count against the same budget. You'll see warnings as you approach any limit.
- Max tiers: Max 5x includes five times and Max 20x twenty times the Pro plan's per-session allowance (What is the Max plan?).
- Weekly reset: on Max, the weekly limit resets at a fixed day and time assigned to your account, regardless of when you start using Claude. Your next reset time is in Settings › Usage.
- What burns usage faster: long, complex conversations, the features and tools you use, the model, and the effort level.
- Check it in the CLI:
/usageshows plan usage bars, activity stats and a breakdown of what counts against your limits (Claude Code docs).
Switching models only helps sometimes. The Claude Code docs say “You've hit your session limit” and “You've hit your weekly limit” are shared across all models, so /model won't restore access. After a model-specific “You've hit your Opus limit” or “Sonnet limit” message, switching to a model outside that family does keep you working.
Watch the API key trap: if ANTHROPIC_API_KEY is set in your environment, Claude Code uses it instead of your subscription and you pay API rates. The monthly API credits that come with Max can't be used in Claude Code or the Claude apps.
Cursor usage limits (monthly pools)
Per Cursor's help center, most plans include two monthly usage pools:
- Cursor Models — Grok 4.7, Grok 4.6, Grok 4.5 and Composer 2.5, tracked separately with “significantly more included usage” (pricing docs).
- Other Models — third-party models, charged at the model provider's price.
Pro ($20/mo), Pro Plus ($60/mo) and Ultra ($200/mo) include both pools; Start, the India-only plan, covers the Cursor Models pool only. Cursor doesn't publish a fixed size for included usage on these pages; its own rough guide is that daily agent users typically spend $60–$100 a month in total usage and power users $200+. Which model you pick changes how fast the pool drains.
- Reset: monthly with your billing cycle; unused usage doesn't roll over. Team members all reset together on the team cycle. The date is on the Spending tab.
- Auto and subagents: Auto bills each request at the routed model's list price, so a third-party route draws from Other Models. Subagents can run a named third-party model even while the picker shows Auto, Grok or Composer.
- BYOK: on individual plans, requests with your own API key don't draw from either pool (your provider bills you). On Teams and Enterprise, BYOK requests still carry the Cursor Token Rate of $0.25 per million tokens, drawn from Other Models.
- Spend limits: with on-demand usage on, you can set a monthly spend limit. When it's reached, AI features stop for that user until the limit is raised or the next cycle starts; enforcement isn't instant, so usage can briefly go over.
Codex usage limits, banked resets and instant resets
Per OpenAI's help center, Work and Codex share the usage allowance included with your plan, and it doesn't correspond to a fixed number of messages or tasks. Depending on the plan, limits apply over a five-hour window and a weekly window, and you need allowance left in both to continue. A new five-hour window starts with your first message after the previous one ends, and you can hit the five-hour limit early even with weekly usage left.
- Pro: Pro 100, Pro 200 and Pro 500 currently have no five-hour limit in Work and Codex. They still have an included allowance with reset times — and “currently” means it can change.
- Rough sizes (OpenAI's estimates, not guarantees): for Plus and Standard Business, OpenAI estimates 5–45 local messages per five-hour window with GPT-6 Astra and 350–3,000 with GPT-6 Luna. Cloud tasks can use more, and weekly limits also apply.
- Check it: Settings › Usage shows remaining allowance and reset times; in an active Codex CLI session, enter
/status(Using Codex with your ChatGPT plan). - Free: auto-review safety checks are free and don't count toward your limits when you're signed in with ChatGPT.
- Mid-task: if you hit a limit during an active turn, Codex can finish that turn, subject to fair-use limits.
Banked, automatic and purchased resets
| Reset type | What it does | Watch out |
|---|---|---|
| Banked reset (promo / referral) | A one-time reset saved to your account. A full banked reset refreshes both the five-hour and weekly windows. Apply it from Settings › Usage (“1 reset available”). | It moves your weekly reset date: use it Tuesday and your next weekly reset is around the following Tuesday, not your old Friday. Not purchased credit, not a permanent increase; it expires. |
| Automatic / global reset | OpenAI occasionally refreshes eligible limits for everyone (e.g. Sep 7, 2026 for Plus, Pro and Business). | Applied directly, can't be saved. Future resets are not guaranteed. |
| Purchased instant reset (eligible Plus and Pro personal accounts) | Restores five-hour and weekly usage as soon as checkout succeeds, from desktop Usage settings or the weekly-limit banner. | Pulls your weekly allowance forward rather than adding extra. Your new week starts with your first Work or Codex request after the reset; applies immediately even if you had usage left. |
Sources: How banked Codex resets work, Paid weekly Work and Codex rate limit resets. OpenAI Support does not reset limits manually; contact it only if usage looks miscounted or a reset didn't apply.
What to do when you hit a limit
| Tool | Free options | Paid options |
|---|---|---|
| Claude Code | Wait for the five-hour or weekly reset (Claude Code v2.1.234+ can wait and continue the interrupted task after the reset); after an Opus- or Sonnet-only limit, switch model family with /model; lower effort and start fresh conversations to stretch usage | Enable usage credits in Settings › Usage (standard API rates, with a monthly spend limit you set); upgrade Pro → Max 5x → Max 20x; or use a Claude Console API account |
| Cursor | Move work to Cursor Models (Grok, Composer), which draw from their own pool; on an individual plan, use your own API key (BYOK) | Enable on-demand usage (same API rates, billed monthly; set a spend limit) or upgrade the plan |
| Codex | Wait for the reset shown in Settings › Usage; apply a banked reset if you have one; pick a lighter model or lower reasoning effort next time (switching models doesn't restore a shared pool) | Buy credits (eligible Plus and Pro), buy an instant reset (eligible Plus and Pro personal accounts), or upgrade |
Picking between them on cost? See Claude Code vs Codex, Claude Code vs Cursor and the subscription columns in the agent config matrix.
We're tracking quota changes
These rules move fast: the OpenAI and Anthropic help pages above were all updated in the past few weeks, and Codex's “Pro has no five-hour limit” is explicitly “currently”. We re-check every source on this page and log real changes to limits, windows and resets on our news page. Quota-change alerts — a short note when one of these limits changes — are what we're building next.
Sources
- Use Claude Code with your Pro or Max plan — Claude Help Center
- What is the Max plan? — Claude Help Center
- Manage usage credits for paid Claude plans — Claude Help Center
- How do usage and length limits work? — Claude Help Center
- Manage costs effectively (/usage, limit messages) — Claude Code docs
- Usage and limits — Cursor Help
- Models & Pricing — Cursor Docs
- Spend limits — Cursor Help
- Using Codex with your ChatGPT plan — OpenAI Help Center
- Managing usage with GPT-6 Astra in Work and Codex — OpenAI Help Center
- How banked Codex resets work — OpenAI Help Center
- Paid weekly Work and Codex rate limit resets — OpenAI Help Center