ChatGPT Work and Codex use one shared agentic allowance when those features are available on the plan. Local messages and cloud chats draw from it, while model, task size, context, reasoning, tools, retrieval, and caching affect how quickly the allowance falls. Regular ChatGPT usage is separate.
What counts against the shared meter
On 2026-10-06, the OpenAI Help Center entry was dated 2026-10-05. The pricing page said ChatGPT Work uses the same pricing, credits, and usage limits as Codex. The Help Center described the shared pool as covering Codex, ChatGPT Work, ChatGPT for Excel, and Workspace Agents when those features are available on the plan. Heavy Work activity can therefore leave less capacity for Codex, and the reverse is also true.
Codex Cloud runs tasks on OpenAI-managed computers. The Help Center says standard Codex Cloud environments have no separate virtual-machine charge at launch, and that model usage counts against normal Codex limits and applicable credits or billing. The pricing page says cloud tasks may use more allowance than local messages. It does not publish a fixed cloud multiplier.
Regular ChatGPT Chat is outside this agentic meter. The Help Center says regular Chat usage is not included in desktop Work and Codex credit usage, and regular Chat Voice has separate limits. Voice work started inside Work or Codex on desktop uses the shared allowance.
What the five-hour and weekly limits mean
The pricing page read on 2026-10-06 listed these estimated local messages per five-hour period for Plus and Standard Business:
Model | Plus | Standard Business |
|---|---|---|
GPT-6 Astra | 5-45 | 5-45 |
GPT-6.1 Sol | 15-160 | 15-160 |
GPT-6 Sol | 15-150 | 15-150 |
GPT-6 Luna | 350-3,000 | 350-3,000 |
These are estimates, not a fixed number of equal-cost messages. OpenAI says consumption depends on the model, task size and complexity, context, reasoning, tools, retrieval, caching, and whether the task runs locally or in the cloud. Cloud tasks may use more than local messages. Weekly limits may also apply.
Speed settings change the rate at which included usage is consumed. Fast mode uses 2.5x the included subscription rate and 2x the purchased-credit or Enterprise pay-as-you-go rate. GPT-6 Astra Ultrafast uses 8x the included rate and 6x the purchased-credit or Enterprise rate. Image generation uses the same general limits as local messages and cloud chats, and uses included limits about 3-5x faster on average.
What credits do after included usage runs out
The plan's included usage is used first. After a limit is reached, supported accounts can draw from a credit balance. Plus and Pro users can buy additional credits without changing plans. Credits are valid for 12 months from purchase, do not roll over, and are generally non-refundable, subject to the exceptions in OpenAI's credit policy. Concurrent tasks can drive a balance negative; later purchases or automatic reloads may offset that negative balance first.
For Business, Enterprise, and Edu workspaces using the credit-based Work and Codex rate card, Standard-mode rates are charged per million tokens:
Model | Input | Cached input | Output |
|---|---|---|---|
GPT-6 Astra | 250 | 25 | 1,250 |
GPT-6.1 Sol | 50 | 2.5 | 250 |
GPT-6 Sol | 50 | 5 | 250 |
GPT-6 Luna | 2.5 | 0.25 | 12.5 |
The applicable plan, workspace agreement, and billing mode determine which rate card applies. Token prices are not a conversion for the included message estimates.
Where the pages disagreed on 2026-10-06
The pages read on 2026-10-06 do not describe every limit the same way.
First, the OpenAI developer pricing page says Pro plans currently have no five-hour limit. The Codex Docs pricing table prints five-hour ranges for Pro 5x and Pro 20x, including 70-700 and 300-3,000 GPT-6 Sol messages. The Help Center also says a full banked reset can refresh five-hour and weekly Codex windows. The pages do not explain whether these statements describe different plan generations or different kinds of reset.
Second, the Help Center says ordinary ChatGPT image-generation limits and banners do not apply to Codex. The developer pricing page says image generation uses the same general limits as local messages and cloud chats and consumes included usage 3-5x faster. Those may be two different meters, but the pages do not define the boundary.
Third, the developer pricing table shows Standard Business with the same estimates as Plus, while the Codex Docs pricing table says Business ($100) uses the Pro 5x estimates. Treat the seat type and the account's usage page as the source of truth before promising a number.
How to check the live allowance
Open https://chatgpt.com/codex/settings/usage for current limits, credit balance, and reset information. During an active Codex CLI session, enter:
/statusThat live account view matters more than a static estimate because the published ranges vary with the model, task, and billing mode.
Sources
- OpenAI Codex pricing and usage limits: https://developers.openai.com/codex/pricing (read 2026-10-06)
- OpenAI Help Center, Using Codex with your ChatGPT plan: https://help.openai.com/en/articles/11369540 (read 2026-10-06)
- Codex Docs pricing table: https://www.codex-docs.com/en/docs/pricing (read 2026-10-06)
- OpenAI Help Center, Using Credits for Flexible Usage in ChatGPT: https://help.openai.com/en/articles/12642688 (read 2026-10-06)
- OpenAI Help Center, ChatGPT rate card for Business, Enterprise, and Edu: https://help.openai.com/en/articles/11481834-chatgpt-rate-card-business-enterpriseedu-credit-based-pricing (read 2026-10-06)
Last verified: 2026-10-06.