Free GPT API · Coming soon
Request a compatible API you can actually connect to Codex
We will not send only a key. You will also get the Base URL, available models, quota, and the exact Client + CLI setup and troubleshooting path on this site.
Leave one email — we send everything when ready
No install, API key, or advance config reading required today.
Used only for API access notices. See the privacy policy.
Why this API is built around Codex first
Generic “OpenAI-compatible” does not guarantee Codex works. We prioritize Responses endpoints, model ids, streaming, and the Client / CLI details that decide whether setup succeeds.
More than chat/completions
Current Codex needs Responses. The rollout will name the correct endpoint so protocol mismatch is visible upfront.
Steps for both Client and CLI
They share ~/.codex, but restart, credentials, and CODEX_HOME can diverge. We verify both surfaces.
Clear quota before you start
Models, quota, rate limits, and rollout timing arrive in the access email, so “free” has concrete boundaries.
After access arrives
- 1
Follow the guide
Wire key / auth.json / config.toml for Client and CLI.
- 2
Switch with the toolkit if you like
Add base_url + key as a provider, Enable, restart the session.
- 3
Try a short task first
Confirm auth and routing before a long session.
FAQ
Can I use it today?
Not yet — leave an email first. We notify when it opens.
Same as an official OpenAI key?
This is our compatible access for running Codex. It is not an OpenAI account.
Can other tools use it?
It follows common call styles; we prioritize Codex Client and CLI.
Next: Client + CLI · Tools