Both cost $20/mo. The difference is what they do.
OpenAI's coding agent: CLI + autonomous cloud PRs
Cognition's AI IDE with the Devin Local agent
Windsurf is now Cognition's Devin Desktop, an IDE billed on Devin plans at $20/mo Pro or $200/mo Max, while Codex is an OpenAI agent bundled into ChatGPT plans that installs into Windsurf as an extension, so the real question is whether you want to pay for the editor or just for the agent.
| Attribute | Codex | Windsurf |
|---|---|---|
| What you are buying | An agent bundled with a ChatGPT plan, no editor of its own | A VS Code fork, renamed from Windsurf to Devin Desktop on 2 June 2026 |
| Price | Free tier, $20/mo Plus, $100/mo Pro 5x, $200/mo Pro 20x | Free, $20/mo Pro, $200/mo Max, Teams from $80/mo with $40/mo full seats |
| Models | OpenAI only: GPT-6 Astra, Sol and Luna, GPT-5.6 during the rollout, plus Codex-Spark on Pro | Cognition SWE models, Claude and GPT, with the Adaptive router choosing per task |
| Do they overlap? | Ships an extension that installs directly into Windsurf | Runs Codex as an extension alongside its own Devin Local agent |
Windsurf is now Cognition's Devin Desktop, an IDE billed on Devin plans at $20/mo Pro or $200/mo Max, while Codex is an OpenAI agent bundled into ChatGPT plans that installs into Windsurf as an extension, so the real question is whether you want to pay for the editor or just for the agent.
Codex at the entry point, because it is included in ChatGPT plans: Free, then $20/mo Plus (and, for a limited time, the $8/mo Go plan). Windsurf, now billed as Devin, is free to try and $20/mo for Pro. The two top individual tiers differ, $200/mo for Devin Max against $500/mo for ChatGPT's top Pro plan. For teams Devin starts at $80/mo with $40/mo full seats, against $25/user/mo for ChatGPT Business billed monthly.
Cognition, the company behind Devin. Windsurf became Devin Desktop on 2 June 2026, delivered as an over-the-air update, and windsurf.com/pricing now redirects to devin.ai/pricing. Cognition's own FAQ says the editor, extensions, keybindings and workflows are unchanged and that existing plans and prices carry over, including legacy Windsurf Enterprise contracts.
Yes. OpenAI publishes the Codex IDE extension for VS Code, Cursor and Windsurf, and it prompts you to sign in with a ChatGPT account or an API key. That means a Windsurf licence and a ChatGPT plan stack rather than compete: Devin Local draws on your Devin quota, Codex draws on your ChatGPT allowance, in the same window.
Both delegate, on different rails. Codex hands long tasks to its cloud lane from the CLI or the extension and returns a reviewable result, and cloud chats draw from the same plan allowance as local ones. Devin Desktop's quota is shared across the desktop IDE, the CLI, cloud Devin sessions and Devin Review, so a single subscription covers local edits and delegated sessions. Choose by which vendor's models you want driving.
OpenAI's coding agent: CLI + autonomous cloud PRs. It sits in the coding agents category, and ships in 1 tested stack here.
Cognition's AI IDE with the Devin Local agent. It sits in the ai ides category, and ships in 1 tested stack here.
They are alternatives at the same step, so most stacks pick one or the other. No aistack.sh stack uses both today, though nothing prevents a custom workflow.
Beyond each other, Codex's closest alternatives are Claude Code, Cursor, GitHub Copilot. Windsurf's closest alternatives are Cursor, Claude Code.
