The key difference
Codex is OpenAI’s coding agent. It runs as an open-source CLI, an IDE extension, a mode in the ChatGPT desktop app, and a cloud agent that runs tasks and opens pull requests. It isn’t sold on its own: it’s included in ChatGPT plans, and it runs OpenAI’s models. OpenAI currently recommends GPT-6.1 Sol for complex coding and GPT-6 Luna for focused tasks.
Cursor is an AI-first editor, a fork of VS Code with an agent, Tab completion, a CLI and cloud agents. Since SpaceX bought it, Cursor is sold through The X Subscription, and its included usage covers Cursor’s own models: Grok 4.7, 4.6 and 4.5, and Composer 2.5.
That makes this comparison less “agent vs editor” than it used to be, and more about which models your subscription pays for:
- Codex = OpenAI models, included in ChatGPT.
- Cursor = Grok and Composer, included. OpenAI models cost API list price on Premium and above.
Pricing in practice
| Tier | Codex (ChatGPT plans) | Cursor |
|---|---|---|
| Free | Free: limited, GPT-6 Luna in the desktop app (subject to rollout) | Free: limited agent and chat requests |
| $8 | Go: lightweight use in the desktop app | Plus: Cursor’s own models only |
| $20 | Plus: Codex on web, CLI, IDE extension and iOS; cloud code review; Slack | — |
| $30 | — | Premium: 4x Plus usage; GPT, Claude, Gemini at API list price |
| $100 | Pro ($100 tier) | Super: 15x Plus usage |
| $200 | Pro ($200 tier) | Ultra: 50x Plus usage |
| $500 | Pro ($500 tier), with Ultrafast mode | — |
| Teams | Business: $25/user ($20 billed yearly), 2+ users | $30, $100 or $200 per seat |
Prices are from the official Codex pricing and Cursor pricing pages, checked October 22, 2026.
Features side by side
| Codex | Cursor | |
|---|---|---|
| Form | CLI, IDE extension, ChatGPT desktop app, cloud agent | VS Code-based editor, CLI, cloud agents |
| Included models | GPT-6.1 Sol, GPT-6 Luna (by plan) | Grok 4.7, 4.6, 4.5; Composer 2.5 |
| Other models | Local and custom providers in the CLI | Claude, GPT, Gemini at API list price (Premium+) |
| Tab completion | Not offered: it’s an agent | Yes |
| Cloud agent | Runs tasks and opens pull requests | Premium and up |
| Code review | Automatic GitHub code review (Plus and up) | Bugbot (Premium and up) |
| IDE support | VS Code, Cursor, Windsurf, JetBrains, Xcode | Cursor itself; JetBrains through ACP, Xcode through Xcode’s MCP bridge (paid plans) |
| Project instructions | AGENTS.md |
Rules |
| Open source | CLI (Apache-2.0) | No |
Using both
This pairing works better than most because the products overlap less than their names suggest:
- Use Cursor as your editor, for Tab completion and quick agent edits on Cursor’s models.
- Install the Codex extension in Cursor for bigger tasks on OpenAI models, paid from your ChatGPT plan.
- Use Codex cloud to hand off work that should come back as a pull request.
Who should pick which
Choose Codex if:
- you already pay for ChatGPT, so Codex costs nothing extra
- you want OpenAI’s models without paying API prices on top of a subscription
- you want an open-source CLI, local models, or Xcode and JetBrains support
Choose Cursor if:
- you want an AI editor with Tab completion and an agent in the same window
- Grok and Composer are good enough for most of your work
- you want the cheapest paid plan ($8 Plus), or you already pay for The X Subscription
Hands-on test
We haven’t run a hands-on test of this pair yet. This comparison is based on each vendor’s documentation and pricing pages, linked on the Codex and Cursor pages. When we add a test, this page will name the task, the tool and model versions, and the date it was run.
