Codex usage limits, plan by plan
Codex limits depend on your ChatGPT plan. On Plus and Business, OpenAI estimates a range of local messages per five-hour window for each model; Pro plans currently have no five-hour limit. Weekly limits may also apply, with no published figure. Your usage dashboard shows the limits that apply to you and when they reset.
The figures OpenAI publishes are estimates, not quotas: two tasks that look alike can use very different amounts, depending on the model, the context, reasoning, tool use and caching. What stays stable is the structure: which plan has a five-hour window, what continues once you reach a limit, and where to see what is left.
Checked on , against OpenAI's Codex pricing page and command reference, listed under Sources. OpenAI adjusts models, plans and limits often: the pages linked below are the reference.
Limits by plan
| Plan | Price | Five-hour window |
|---|---|---|
| Free | $0 | No figure published; for quick coding tasks |
| Go | $8 a month | No figure published; for lightweight coding tasks |
| Plus | $20 a month | A range of local messages per five hours, by model (table below) |
| Pro | From $100 a month ($100, $200 or $500) | None: “Pro plans currently have no five-hour limit” |
| Business | $20 per user a month billed yearly, two users or more; $25 billed monthly | Standard Business: the same ranges as Plus |
Source: learn.chatgpt.com/docs/pricing, checked on . Prices in US dollars. Enterprise and Edu plans with flexible pricing have no fixed rate limits; without it, they have the same per-seat limits as Plus for most features.
| Model | Local messages per five hours, Plus and Standard Business |
|---|---|
| GPT-6.1 Sol | 15–160 |
| GPT-6 Sol | 15–150 |
| GPT-6 Astra | 5–45 |
| GPT-6 Luna | 350–3,000 |
Source: the same page, same date. OpenAI calls these estimates rather than fixed limits; cloud tasks may use more of the allowance than local messages.
The weekly window
OpenAI's pricing page says that weekly limits “may also apply” and gives no figure for them. In practice, it is often the window that matters: on the Codex plans we measured in September 2026, only the seven-day window was published. A five-hour window with room left does not help once the week is used up, so read both before you start a long task.
Some features draw on the same allowance faster: Fast mode uses included limits 2.5 times as fast as Standard, and image generation three to five times as fast on average. How Codex and Claude Code compare on these windows is set out in Codex vs Claude Code.
Is there a 5-hour limit?
It depends on the plan. Plus and Standard Business are described with a five-hour window; Pro has none at present. Free and Go have no published figure. Each account's dashboard shows which windows apply to it and when each resets, and you can check your Codex usage there, in the ChatGPT desktop app or in the CLI.
When you hit the limit
- The current turn finishes. If you reach a limit during an active turn, the agent can continue working on that turn, subject to fair use limits.
- Credits. Plus and Pro users can buy additional credits to keep working without changing plan; Business, Edu and Enterprise plans with flexible pricing buy workspace credits.
- An API key. Anyone can run extra local chats with an API key, charged at standard API rates.
Banked resets
Some accounts also hold banked resets: a reset saved to the account until it is used or
expires. In the Codex CLI, /usage is where you view your token usage or use a
rate-limit reset. How many you have, until when, and when Quotum spends one:
Codex resets.
Several ChatGPT accounts at once
The dashboard and the CLI each show the account you are signed in to. Claude has a different pair of windows, set out in the guide to check your Claude usage.
Quotum shows each Codex account in your Mac menu bar: the windows its plan publishes, five hours, seven days or both, the time each resets, the resets available, and how long ago the figure was measured. Details: the Codex usage tracker.
The trial is free, lasts fourteen days and covers the first two lines of the panel. Launch price: €4.99, then €9.99. Reading a provider's quota from a third-party app is not something the provider provides for: the pricing page explains the risk before any purchase.
Sources
Consulted on .
- OpenAI, Codex pricing:
learn.chatgpt.com/docs/pricing - OpenAI, Codex CLI commands:
learn.chatgpt.com/docs/developer-commands
FAQ
Does Codex have a 5-hour usage limit?
It depends on the plan. OpenAI describes a five-hour window on Plus and Standard Business, and says that Pro plans currently have no five-hour limit. Weekly limits may also apply.
Does the Codex usage limit reset every day?
No daily reset is published. Codex counts usage in a five-hour window on some plans and in a weekly window, and each resets at the time your usage dashboard shows.
How much can I use Codex for free?
The Free plan includes Codex for quick coding tasks, but OpenAI publishes no figure for its limits. Your usage dashboard shows what applies to your account.
What happens when I hit the Codex usage limit?
A turn already running can finish, subject to fair use limits. After that, Plus and Pro users can buy credits, and anyone can run extra local chats with an API key at standard API rates.
Is Codex really unlimited?
Only Enterprise and Edu plans with flexible pricing have no fixed rate limits; their usage scales with credits. Every other plan has limits.
How do I get more Codex usage?
Buy credits on Plus or Pro, move to a higher plan, use a banked reset if your account has one, or run local chats with an API key, billed at API rates.