ChatGPT Plus gives you two Codex limits: a 5-hour limit and a separate weekly limit. There isn't a fixed number of Codex tasks you get. To get a useful sense of your own limit, check Settings > Usage, run one normal Codex task that's similar to the work you usually do, then check Usage again to see how much it used. Bigger tasks can use much more than smaller ones.
What Are the ChatGPT Plus Codex Limits?
The $20/month Plus plan includes Codex access across the CLI, IDE extensions, desktop app, and cloud tasks. Those surfaces share the same Codex allowance. Regular ChatGPT messages, file uploads, and image generation use separate limits.
Consumer plan base price
Short-term Codex usage window
Separate from the 5-hour limit
Test a normal task and check Usage
How to Check Your Plus Quota
Before a normal task, check your 5-hour and weekly usage in Settings > Usage or by running /status. Run the task, then check again. The change gives you a rough baseline for how much similar work uses. If you want a practical range, repeat once with a heavier task.
Current ChatGPT Plus Codex Limits
OpenAI also publishes estimated local-message ranges by model. They're useful reference data, but they do not tell you how many Codex tasks Plus includes. Actual usage varies by the task, model, settings, and whether the work runs locally or in the cloud. OpenAI also does not publish one universal weekly cap.
OpenAI estimates 15 to 160 local GPT-6.1 Sol messages per five-hour window for Plus and Business Standard. These are estimates, not guaranteed counts. See GPT-6.1 Sol availability and usage limits for current access, model selection, and credit rates.
| Window | What you get | What happens at the limit | Source |
|---|---|---|---|
| 5-Hour Window — GPT-6 Astra | 5–45 local messages (official estimate) | Actual usage varies; check your account's remaining allowance and reset time | Official (estimate) |
| 5-Hour Window — GPT-6.1 Sol | 15–160 local messages (official estimate) | Actual usage varies; check your account's remaining allowance and reset time | Official (estimate) |
| 5-Hour Window — GPT-6 Sol | 15–150 local messages (official estimate) | Actual usage varies; check your account's remaining allowance and reset time | Official (estimate) |
| 5-Hour Window — GPT-6 Luna | 350–3,000 local messages (official estimate) | Actual usage varies; check your account's remaining allowance and reset time | Official (estimate) |
| 5-Hour Window — GPT-5.6 Sol | 10–100 local messages (official estimate) | Actual usage varies; check your account's remaining allowance and reset time | Official (estimate) |
| 5-Hour Window — GPT-5.6 Terra | 25–200 local messages (official estimate) | Actual usage varies; check your account's remaining allowance and reset time | Official (estimate) |
| 5-Hour Window — GPT-5.6 Luna | 250–2,000 local messages (official estimate) | Actual usage varies; check your account's remaining allowance and reset time | Official (estimate) |
| 5-Hour Window — GPT-5.5 | 15–80 local messages (official estimate) | Actual usage varies; check your account's remaining allowance and reset time | Official (estimate) |
| Weekly Allowance | Not published by OpenAI | Check reset time and available credits or reset options | Not published |
| Cloud Task Allowance | Shared Work and Codex allowance | Cloud tasks can use more allowance than local messages | Official |
| Subscription Token Ceiling | No universal cap published | Model context size is separate from subscription allowance | Not published |
To compare against Free, Pro, and Business tiers, see the complete codex usage limits table.
When Does Your Plus Quota Reset?
ChatGPT Plus Codex resets run on dual short-term and weekly schedules, visible in Settings > Usage and CLI /status. Short-term capacity refreshes over a rolling 5-hour window; the weekly window resets on an account-specific schedule that can be moved forward with a purchased instant reset.
Short-Term Capacity Window
OpenAI tracks short-term usage on a rolling 5-hour window rather than a fixed daily reset. Current reset timing is shown live in your CLI and account settings.
Weekly Reset Schedule
The underlying weekly anchor schedule is unpublished. Purchasing a paid instant reset starts a new weekly period immediately.
For a deeper look at reset timing, see when your Codex usage resets.
What Happens When Plus Hits the Limit?
When you hit a ChatGPT Plus limit, Codex can continue working on the active turn, subject to fair-use limits, before pausing subsequent prompts. You'll see a limit notice with the reset time and options to wait, apply a banked reset, add credits, or upgrade. Selected personal Plus accounts may also see a separate Luna Reserve allowance for GPT-5.6 Luna after regular usage is exhausted.
If you're currently blocked, see why am I rate limited and what happens now for recovery options.
How to Avoid Exhausting Plus Limits
You can avoid exhausting your ChatGPT Plus quota by scoping terminal prompts, ignoring unnecessary build directories, and reserving cloud tasks for reviews that need them. Continuous multi-file agent workflows tend to consume Plus allocations fastest.
Give Codex the whole job at once instead of drip-feeding it a conversation. One well-specified turn costs far less than six small ones.
Don't let lockfiles, build artifacts, or minified bundles get pulled in. They cost tokens and tell the model nothing.
Add pay-as-you-go credits via Settings > Usage, or authenticate the Codex CLI with an OpenAI API key to switch to metered OpenAI Platform API billing.
If you routinely exceed Plus limits, OpenAI offers three Pro tiers: $100/mo (published 5x tier), $200/mo (current allowance unquantified), and $500/mo (highest included Pro usage and Astra Ultrafast). Eligible Pro $200 subscribers keep the previous allowance through October 29, 2026. For an economic breakdown of when upgrading makes financial sense, see our Codex pricing and plans guide.
How to Keep Plus Usage Visible While You Work
Codex Usage is a free, local tool that shows your 5-hour and weekly remaining percentages and reset times in your terminal or a local browser dashboard — without a separate account or API key.
Install Codex Usage Locally
Runs locally · No cloud account or telemetry
curl -fsSL https://www.codexusage.dev/install.sh | shFrequently Asked Questions
Does ChatGPT web chat share the same limit as the Codex CLI?
No. OpenAI separates everyday ChatGPT conversational web chat from Codex. Standard web messages, file uploads, and image generation run on independent meters that don't consume your Codex allowance. Codex shares its agentic pool across the CLI, IDE extensions, desktop app, and cloud tasks.
Can I buy additional Codex usage on ChatGPT Plus?
OpenAI offers pay-as-you-go credits and paid instant weekly resets in Settings > Usage for Plus subscribers when plan capacity is exhausted. Alternatively, authenticating the Codex CLI with an OpenAI API key switches billing to the metered OpenAI Platform API instead of your subscription quota.
Do failed CLI executions consume my Plus allowance?
Yes. Allowance is consumed whenever the model processes prompt tokens and generates a response, regardless of whether a subsequent local command succeeds or errors.
When should a developer upgrade from Plus ($20/mo) to Pro?
Consider Pro if you regularly hit the Plus 5-hour or weekly window during normal work. OpenAI offers Pro $100 ($100/mo, published 5x tier), Pro $200 ($200/mo, current multiplier unpublished), and Pro $500 ($500/mo, highest included Pro usage and Astra Ultrafast). Eligible Pro $200 subscribers keep the previous allowance through October 29, 2026. The exact weekly allowance for each tier isn't published.