OpenAI Codex

OpenAI Codex is a software engineering agent rather than an autocomplete tool. You hand it a task, such as tracking down a bug, refactoring a module, adding a feature or writing tests, and it works through the codebase, runs commands and returns changes for you to review. The same agent is available in the ChatGPT desktop, web and mobile apps, in the terminal through the Codex CLI, in an IDE extension and in Codex cloud, all connected through your ChatGPT account.

As of October 2026, OpenAI's Codex pricing page lists GPT-6.1 Sol, GPT-6 Sol, GPT-6 Luna and GPT-6 Astra as the models used in Codex. GPT-5.5 was retired from Codex for ChatGPT sign-in on October 14, 2026. In June 2026 OpenAI said Codex had more than 5 million weekly active users, and JetBrains research published in August 2026 put Codex's work adoption at 16% in May to July 2026, up from 3% in January.

Codex can run several agents in parallel, use computer and browser tools to check its own work, and keep working on cloud tasks after you close your laptop, with reusable cloud environments for each repository. It can automatically review pull requests, surface findings and answer questions about them. It connects to GitHub, GitLab (beta), Slack and Linear and can be extended with plugins. The Codex CLI is open source under the Apache-2.0 license and installs on macOS, Linux and Windows, and a Codex SDK and the codex exec command support scripting.

Codex is included in ChatGPT plans. Free ($0) and Go ($8 a month) include lightweight Codex use in the desktop app with GPT-6 Luna, subject to rollout. Plus ($20 a month) adds Codex on the web, in the CLI, the IDE extension and on iOS, with GPT-6.1 Sol and GPT-6 Luna. Pro costs $100, $200 or $500 a month and currently has no five-hour usage limit. Business is $20 per user per month billed annually for 2 or more users, or $25 per user billed monthly, and Enterprise and Edu are priced through sales. Plus, Pro and Business users can buy extra credits, and Codex can also run with an API key at standard API rates.

The main limitation is usage. On Plus and Business, Codex is capped per five-hour window, and heavier models use the allowance much faster: OpenAI estimates roughly 15 to 160 local messages per five hours with GPT-6.1 Sol, against 350 to 3,000 with GPT-6 Luna. Free and Go only get the lightweight Luna model in the desktop app. Codex works only with OpenAI models, and as with any coding agent, its changes and its shell and repository access need careful review and permission settings.

Codex suits developers and teams who already pay for ChatGPT and want one agent across the terminal, the editor and the cloud, and teams that want AI review on every pull request. Developers who prefer Anthropic's models or want to switch between vendors should also look at Claude Code, Cursor or GitHub Copilot.