Codex Pricing (2026): Plans, API Costs & Limits

Quick answer

ChatGPT Plus includes Codex at $20/month in the US price list. Sign in with ChatGPT for plan access and cloud features. For supported local workflows, an API key bills the Platform account by model and usage. Check your account’s allowance and required features before choosing.

Independent guide. Updated . Facts last verified .

Choose a Codex billing option in 30 seconds

  • Occasional personal coding. Try your existing ChatGPT access on one repository. Check model and feature access before paying for a higher tier.
  • Daily personal coding. Start with the lowest plan that supports your workflow. Compare Pro tiers if limits repeatedly interrupt reviewed work.
  • Team repositories. Confirm workspace access and admin controls. Ask the workspace owner which usage and extra credits the team pays for.
  • Private CI or scripts. Use API-key billing for supported local automation. Confirm model access and estimate retries before scheduling jobs. Codex cloud requires ChatGPT sign-in.

Key facts

ScopeOpenAI Codex
Last verified
DecisionCompare subscription access and metered API billing separately.
Official documentation

OpenAI plan pricing and limits

OpenAI authentication guide

OpenAI API pricing

What are the published monthly options?

Prices checked . These are monthly USD list prices before tax. Allowances differ by plan.

PlanUSD/monthUsage and billing
ChatGPT Plus$20Shared ChatGPT Work and Codex allowance. Usage depends on the model and task.
ChatGPT Pro$100Check the allowance and feature access for this tier in your account.
ChatGPT Pro$200Check the allowance and feature access for this tier in your account.
ChatGPT Pro$500Check the allowance and feature access for this tier in your account.

Which ChatGPT plans include Codex?

OpenAI lists limited desktop access on Free and Go, with Go listed at $8/month and access subject to rollout. Plus and Pro include broader Codex access. Business, Enterprise and Edu users should confirm workspace features and administrator controls before choosing a plan.

Does every plan have the same five-hour limit?

OpenAI publishes five-hour estimates for Plus and Standard Business. Pro currently has no five-hour limit, though other plan restrictions can apply. Model choice, task size and cloud work affect consumption. Use the dashboard for your actual allowance and reset time.

What does API authentication cost?

API-key usage is metered by the selected model. Input, cached input, cache writes and output can have different rates. Long context, service tiers and tool charges can change the invoice. The worked examples use standard short-context uncached input and output.

Which account pays for Codex?

  • ChatGPT sign-in. Usage follows the active personal plan or workspace. Check the plan allowance and any purchased credits in that account.
  • API key. Model usage and applicable API tool charges go to the OpenAI Platform billing account associated with the key.
  • Enterprise automation. Ask your workspace admin whether Codex access tokens are enabled for trusted local workflows. They use workspace entitlements.

See the authentication guide for access by work surface and sign-in method.

Estimate Codex API cost for your workload

For gpt-6.1-sol at standard short-context rates, multiply millions of uncached input tokens by $2 and millions of output tokens by $10, then add the two amounts. These hypothetical monthly workloads exclude caching, long context, service tiers, tool charges and taxes.

  • Small local trial. 0.5M input + 0.1M output = $2.00.
  • Recurring repository work. 5M input + 1M output = $20.00.
  • Private CI workload. 20M input + 4M output = $80.00.

How to check your Codex quota and billing

  1. Confirm the active account and workspace. In the CLI, run codex login status to check the sign-in method.
  2. For ChatGPT sign-in, open the Codex usage dashboard and record the remaining allowance, reset time and affected model or workspace.
  3. For API-key use, check the correct project in the Platform usage dashboard. Separate model and tool charges before comparing them with your estimate.

Next steps

Frequently asked questions

Does this subscription include unlimited tokens?

No fixed unlimited token entitlement is assumed here. Read the official plan rules and your account's current usage screen; exact subscription token allowances are not publicly specified as a stable conversion.

Can I estimate my API invoice from hours alone?

Hours are only an input assumption. You also need actual input and output tokens, model, caching, context length and any tools or service-tier charges. Compare the estimate with recorded billed usage.

Where should I compare other tools?

Use the pricing hub and cost calculator for a common assumption set. Check each provider's source again before purchasing because plans, limits and regional taxes can change.

Official sources

Checked . Next scheduled fact review .

On 2026-11-07, we checked public US plan prices, usage windows, supported authentication methods and the model rates used in the examples. Account-specific quotas and regional checkout prices require a check in your own account. See our editorial policy.