If you want to use OpenAI Codex for coding, the main cost is your ChatGPT plan. Codex is included with ChatGPT plans, including Free, but each plan has different usage limits. Codex Cloud, which runs coding tasks on OpenAI-managed computers, is available only on eligible paid plans and depends on account and workspace settings.
There is no separate Codex subscription. You can also use Codex with an OpenAI API key, which bills separately based on API usage. The right option depends on whether you prefer a predictable subscription allowance or pay-as-you-go API charges.
What ChatGPT plans cost
The listed monthly prices for Plus and Pro are:
| ChatGPT plan | Listed price | What to know about Codex |
|---|---|---|
| Free | Free | Codex is included with restricted usage. Codex Cloud is not included. |
| Go | Varies by market | Codex is included with plan limits. Codex Cloud is not included. |
| Plus | $20 per month | Includes Codex access, with lower usage limits than Pro. |
| Pro | $100, $200, or $500 per month | Higher usage levels; the available allowance depends on the tier. |
| Business, Enterprise, Healthcare, and Education | Varies by plan | Codex access and billing depend on the plan and workspace settings. |
OpenAI lists the plan features and current limits on its Codex pricing page. Its plan access information says Codex Cloud is available to eligible Plus, Pro, Business, Enterprise, Healthcare, and Education accounts. Free and Go plans do not include Codex Cloud. Workspace settings and rollout eligibility can also affect access.
For a comparison between Plus and Pro, VentureBeat reported that the $100 Pro tier provides five times the Codex usage limits of Plus. The fivefold figure compares included usage limits; the number of coding tasks varies with the work and the model.
How Codex usage limits work
Your plan does not include a fixed number of messages. A short coding question uses less capacity than a task that reads a large codebase, runs tools, or requires several rounds of reasoning. The model, task complexity, context, and caching all affect how much usage a task consumes, according to OpenAI’s pricing details.
Local Codex chats and cloud tasks draw from the same plan allowance. Other agent features also use this allowance. OpenAI lists ChatGPT for Excel as one example for Plus and Pro. Image generation and desktop voice also use the general allowance, Using them leaves less capacity for coding. Fast and Ultrafast modes consume included usage more quickly than Standard mode.
When you approach a limit, check the usage dashboard or enter /status during an active Codex CLI session. If your account offers additional credits, you can purchase them to continue after reaching a plan limit. For work billed through an API key, usage is charged at standard API rates rather than through the ChatGPT plan’s allowance.
Choosing between a plan and API billing
A ChatGPT plan suits people who want Codex access bundled with other ChatGPT features and prefer a recurring subscription. Plus is the lower-cost paid option; Pro offers higher usage for heavier coding. Choose a plan based on how often you use Codex. Check the current allowance instead of assuming the plan covers a set number of tasks.
API-key billing suits people who want direct usage-based charges. However, Codex CLI and the IDE extension do not provide built-in cost tracking for API spending, and the cost-tracking analysis notes that native views do not show costs for each session or project. Set a spending limit in the billing controls and review usage regularly, especially when Codex runs long tasks or uses premium models.