OpenAI Codex
Coding agent by OpenAI
OpenAI's coding agent, available as an open-source terminal CLI, an IDE extension, a mode in the ChatGPT desktop app, and a cloud agent that runs tasks and opens pull requests. Usage is included in ChatGPT plans or billed per token with an API key.
Key facts
- Type
- Coding agent · CLI, IDE extension, Desktop app, Cloud agent, Web
- Pricing
- Free tier, paid from $8/mo
- Free tier
- ChatGPT Free and Go include limited Codex in the desktop app (GPT-6 Luna at standard speed, subject to rollout), without Codex Cloud. The CLI can also run on an OpenAI API key billed per token, or on local models through Ollama or LM Studio.
- License
- Apache-2.0
- Models
- GPT, Local models via Ollama or LM Studio, Custom OpenAI-compatible providers
- Platforms
- macOS, Windows, Linux, Web
- Bring your own API key
- Yes
- Local models
- Yes
- MCP support
- Yes
- Self-hostable
- No
- GitHub stars
- 130,413
- Last commit
- Oct 22, 2026
OpenAI Codex pricing
Free
$0/month
Limited Codex in the desktop app with GPT-6 Luna at standard speed, subject to rollout; no Codex Cloud
Go
$8/month
Codex in the desktop app with GPT-6 Luna at standard speed, subject to rollout; no Codex Cloud. No annual billing
Plus
$20/month
Codex on web, CLI, IDE extension and iOS with GPT-6.1 Sol and GPT-6 Luna; automatic code review and Slack; buy credits past the limits. No annual billing
Pro 100
$100/month
Higher limits than Plus; no 5-hour limit at present (weekly limits may apply). No annual billing
Pro 200
$200/month
More usage than Pro 100. New subscribers get a lower allowance than existing ones, who keep theirs through Oct 29, 2026
Pro 500
$500/month
Highest usage; the only individual plan with Ultrafast mode
Business Standard seat
$25/month
$20/month billed yearly
Minimum 2 seats. Pooled credits for extra usage; admin controls for models, speed and cloud access
Business Premium seat
$125/month
$100/month billed yearly
Higher limits than a Standard seat
Enterprise & Edu
Custom
Contact sales. Plus-level limits per seat, or no fixed limits with flexible pricing and a shared credit pool
Full OpenAI Codex pricing guide: usage limits, API costs and which plan to pick →
Best for
- ChatGPT subscribers who want a coding agent without a separate subscription
- Developers who want an open-source CLI agent they can inspect or extend
- Running agents in CI with an API key
Pros
- CLI is open source under Apache-2.0
- Included in ChatGPT plans from Free upward
- Supports local models (Ollama, LM Studio) and custom providers in the CLI
- Works in VS Code, Cursor, Windsurf, JetBrains, and Xcode
Cons
- Cloud features (GitHub code review, Slack) need a ChatGPT plan, not an API key
- Usage limits vary by model and task and are hard to predict
- Hosted plans center on OpenAI models
How OpenAI Codex compares
Claude Code vs OpenAI Codex
Both cost $20/month to start and do the same core job, so pick by ecosystem.
Updated Oct 22, 2026
OpenAI Codex vs Cursor
If you want OpenAI's models, pick Codex: it's included in ChatGPT plans (limited on Free and Go, full from Plus at $20/month), while Cursor's plans include only its own Grok and Composer models and bill GPT at API list price from the $30 Premium plan.
Updated Oct 22, 2026
Sources (14)
- pricingPlans: https://learn.chatgpt.com/docs/pricing
- license: https://github.com/openai/codex/blob/main/LICENSE
- platforms: https://github.com/openai/codex/blob/main/README.md
- platforms: https://learn.chatgpt.com/docs/app
- platforms: https://learn.chatgpt.com/docs/linux/linux-app
- modelsSupported: https://learn.chatgpt.com/docs/config-file/config-advanced
- modelsSupported: https://learn.chatgpt.com/docs/models
- status: https://learn.chatgpt.com/docs/codex/cli
- pricingPlans: https://help.openai.com/en/articles/11369540
- pricingPlans: https://help.openai.com/en/articles/9793128-about-chatgpt-pro-tiers
- pricingPlans: https://help.openai.com/en/articles/8792828
- pricingPlans: https://help.openai.com/en/articles/12642688
- pricingPlans: https://developers.openai.com/api/docs/pricing
- pricingPlans: https://learn.chatgpt.com/docs/agent-configuration/speed