Direct answer
Codex usage is included in ChatGPT Free, Go, Plus, Pro, Business, and Enterprise plans, but the allowance differs. Free and Go provide limited trial access. Plus costs $20 monthly and supports focused weekly coding sessions. Pro costs $100 monthly for 5x Plus usage or $200 for 20x. Additional credits and API-key billing can extend eligible accounts. Check the Codex usage dashboard because task size, model, context, and shared agentic work all affect consumption.
Key takeaways
- Codex is available on the Free plan, but access is a limited trial rather than an unlimited free coding service.
- Plus is the $20 individual entry for expanded Codex use; Pro adds 5x or 20x the Plus allowance.
- Local messages and cloud tasks can share a rolling five-hour window, and weekly limits may also apply.
- Codex, ChatGPT Work, ChatGPT for Excel, and Workspace Agents can draw from one agentic usage pool.
- Credits continue eligible usage after an included limit and are consumed according to model and token activity.
- The account's usage dashboard is more reliable than a generic estimate for deciding when to change plans.
Codex plans and current prices
There is no separate individual Codex subscription on the current pricing page. Individual access comes with a ChatGPT plan. Prices below are the current US list prices checked on 10 August 2026. Tax, app-store billing, localization, contracts, and later changes can alter the amount paid.
- Free
- $0 with limited trial access to Codex capabilities.
- Go
- $8 monthly in the US, with limited Codex trial access and a larger general ChatGPT allowance than Free.
- Plus
- $20 monthly with expanded Codex usage and access to the Codex desktop app.
- Pro 5x
- $100 monthly with five times the Plus usage allowance.
- Pro 20x
- $200 monthly with twenty times the Plus usage allowance.
- Business
- the standard annual plan is listed at $20 per seat per month equivalent and includes Codex limits.
- Enterprise and Edu
- access, credit structure, limits, and contract pricing depend on workspace configuration.
- API key
- local tasks can use separately billed API capacity at current model token rates.
Is Codex free?
Yes, Codex free access exists as a limited trial inside the ChatGPT Free plan. The current pricing page does not publish a fixed Free message count. Availability can also depend on the model, feature, region, and current account. Use Free to test a small, bounded task. Choose a paid plan only after you can compare the output and observed usage with work you already understand.
How Codex pricing works
Codex pricing has three layers. A ChatGPT plan supplies included usage. Credits can extend eligible use after that allowance. An API key creates a separate usage-based bill for supported local tasks. Do not assume that paying for ChatGPT also covers API use. Confirm the active sign-in method and account before starting a long session.
How Codex usage limits work
Codex limits are not one permanent message number. Local messages and cloud tasks share a rolling five-hour window where those features are available. Additional weekly Codex usage limits may apply. Repository size, context, model, and task difficulty can change consumption. Parallel work, image generation, code review, and fast mode can also use more of the allowance.
Published Plus ranges
The current pricing table gives Plus ranges per five-hour window, not guaranteed task counts. It lists 15 to 90 local messages for GPT-5.6 Sol, 20 to 110 for Terra, and 50 to 280 for Luna. The ranges are wide because messages use different amounts of context and output. Pro is marketed as 5x or 20x the Plus allowance. Check the live table before purchasing.
One shared agentic pool
Several OpenAI products can share one agentic allowance and credit pool. These include Codex, ChatGPT Work, ChatGPT for Excel, and Workspace Agents when the plan includes them. A user may therefore see Codex capacity change after work in another agentic product. Ordinary Chat Voice uses separate caps. Tasks started through Voice can draw from the shared pool.
Check the Codex usage dashboard
Open Codex Settings, then Usage, to view current limits, recent activity, and available credits. During an active Codex CLI session, run /status to see the remaining limit information available to that session. A workspace may also expose a planning estimate in dollars. OpenAI says that estimate is not an invoice. Use the billing or credit record for actual charges.
- Five-hour window
- short-term capacity shared by applicable local and cloud activity.
- Weekly window
- a separate longer allowance that can stop work even after a five-hour reset.
- Credit balance
- purchased capacity available after included usage where the plan supports it.
- Recent usage
- account activity that helps explain a change in the remaining allowance.
- Thread estimate
- a planning figure that may appear when an organization enables cost visibility.
What happens when you reach a Codex limit
An active turn can finish after the account reaches a limit, subject to fair-use controls. Then you may wait for a reset, apply an eligible banked reset, buy credits, or change plans. You can also use an API key for a separately billed local task. The exact options appear in the account's usage page or limit banner.
Token-based credit costs
Most customers have moved from legacy per-message estimates to a token-based rate card. Credits remain the purchased unit. Consumption maps to input, cached input, and output tokens for the selected model. Output-heavy work and fast mode generally use more credits. A small subset of Enterprise accounts can remain on the legacy card until migration.
Team cost controls
New Business workspaces cannot add pay-as-you-go Codex-only seats after 24 June 2026. Existing eligible Business seats can continue under the documented rules. Workspace owners can buy credits, set automatic reload, assign seat or user limits, and review billing controls. Enterprise terms can support flexible credit use and organization-specific access. Confirm the contract and workspace configuration before estimating team cost.
Choose a Codex plan
- Free or Go
- test whether Codex fits a small real task before paying for capacity.
- Plus
- support a few focused sessions each week when interruptions are acceptable.
- Pro 5x
- handle longer or more frequent individual sessions when Plus limits are a repeated constraint.
- Pro 20x
- fit sustained high-intensity individual use only after dashboard history shows the demand.
- Business
- combine ChatGPT workspace features, administration, and included Codex limits for a team.
- Enterprise or API billing
- use negotiated controls or direct token accounting for large, variable, or automated workloads.
Make Codex usage last longer
- Use a smaller or less expensive model for routine work when quality remains acceptable.
- Remove unrelated context and start a fresh task when the goal changes.
- Keep AGENTS.md instructions scoped and nest them near the code that needs them.
- Disable MCP servers and plugins that the current task does not need.
- Avoid parallel agents until the task can justify the added consumption.
- Review the usage dashboard after representative work before upgrading or enabling auto top-up.
Install Codex and verify the first repository session
Install Better Design only when the task needs design guidance
Frequently asked questions
The current individual choices range from limited Free access to Plus at $20 monthly and Pro at $100 or $200 monthly. Team and API costs depend on seats, credits, models, tokens, and workspace terms.
The ChatGPT Free plan currently includes limited trial access to Codex. It does not promise unlimited use or a permanent message count. API-key usage is billed separately.
Open Codex Settings and select Usage. In Codex CLI, use /status for the current session. The usage page or limit banner shows the account-specific options after a limit.