OpenAI
Codex
Coding agent included with ChatGPT plans.
OpenAI's Codex pricing page says Codex is included in ChatGPT Free, Go, Plus, Pro, Business, Edu, and Enterprise. ChatGPT Work and Codex share usage. An API key runs Codex in the CLI, SDK, or IDE extension without the cloud features, and you pay API token rates.
Pricing
As of 15 Oct 2026.
| Tier | Price | Billing | What the page states |
|---|---|---|---|
| Free | $0/month | Free |
|
| Go | $8/month | Monthly |
|
| Plus | $20/month | Monthly |
|
| Pro | $100, $200, or $500/month | Monthly. Ultrafast mode is named on Pro $500. |
|
| Business | $25/user/month when billed monthly | 2 or more users. Annual billing is mentioned. Annual price not extracted. |
|
| Enterprise and Edu | Unknown | Sales or agreement. Flexible pricing is described as credit-based with no fixed rate limits. |
|
| API key | API token rates | Pay for tokens |
|
- Free is $0. Go is $8/month. Plus is $20/month. Pro is offered at $100, $200, or $500 per month.
- A business line on the page says $25 per user per month when billed monthly, for 2 or more users, and that annual billing exists. The annual per-user dollar amount was not in the text extracted on 15 October 2026.
- Enterprise and Edu prices are not given as a public seat rate on the page.
- Voice in Desktop uses the Codex usage budget at $0.05 per minute.
- GPT-5.5 retires from Codex with ChatGPT sign-in on 14 October 2026 at noon PDT. API-key sessions are not affected.
- The long feature matrix on the page is rendered with icons. This pass did not capture every icon, so those rows are not turned into yes or no cells unless the surrounding prose states them.
Feature matrix
| Price | |
| Published from | Free, or Go $8/mosource Codex is included on Free. Go is the lowest paid plan at $8/month. |
|---|---|
| Free tier | Yessource Free is $0 and includes Codex for quick tasks, with the desktop model note above. |
| Product | |
| Open source | Nosource Included with ChatGPT plans or billed as API usage. No open-source license is stated on the pricing page. |
| Agent edits code | Yessource The page describes local and cloud coding tasks, code review, and an agent that can finish an in-progress turn after a limit is hit. |
| Inline completion | Unknownsource Composer predictions are a row in the feature table. The icons were not captured, so inline completion is not marked yes or no. |
| Surfaces | |
| CLI | Plus and API keysource Plus lists the CLI. The API-key card lists the CLI. Free and Go cards do not name the CLI. |
| Models | |
| Bring your own key | Yessource API key access is a documented way to run Codex, billed at API pricing, without cloud features. |
| MCP | Unknownsource MCP is a row on the feature table. Checkmarks were not captured in this pass, so plan availability is unknown here. |
| Surfaces | |
| Cloud agent | Plus, not API keysource Plus lists cloud integrations. The API-key card says there are no cloud features. Free and Go cards do not mention cloud. |
| Price | |
| Team or enterprise admin | Yessource Business prose names a workspace, SAML SSO, and MFA. Enterprise prose names SCIM, RBAC, and audit logs. |
Models
Plus names GPT-6.1 Sol and GPT-6 Luna. Free and Go name GPT-6 Luna at Standard speed in the desktop app, subject to rollout. A usage table estimates local messages per five hours for Plus and Standard Business: GPT-6 Astra 5–45, GPT-6.1 Sol 15–160, GPT-6 Sol 15–150, GPT-6 Luna 350–3,000. The page says Pro plans currently have no five-hour limit. Credit rates per million tokens are also published for those models and for GPT-5.6 Sol, Terra, and Luna.
Platforms
- Web, CLI, IDE extension, and iOS. Named on the Plus card. The API-key card names CLI, SDK, and IDE extension, and excludes cloud features. source
- Desktop app. Free and Go cards name GPT-6 Luna in the desktop app, subject to rollout. source