model is not supported when using Codex with a ChatGPT account.
The message
model is not supported when using Codex with a ChatGPT account.What it means
OpenAI's server refused the model name Codex sent for your ChatGPT plan. The model has been retired from ChatGPT sign-in, needs a higher plan, or hasn't reached your account yet.
What to do
Open /model and pick one from the list, or start with codex -m and a current model. Then delete the old name from the model line in ~/.codex/config.toml so it doesn't come back.
You send a prompt and Codex prints the server's refusal instead of an answer:
■ {"type":"error","status":400,"error":{"type":"invalid_request_error","message":"The 'gpt-5.3-codex' model is not supported when using Codex with a ChatGPT account."}}The desktop app shows only the sentence inside: The 'gpt-6-astra' model is not supported when using Codex with a ChatGPT account. It's the same refusal. The wording doesn't live in the Codex source (we searched the rust-v0.157.1 tag and found it only in a test), so it's OpenAI's server deciding, based on which plan your ChatGPT sign-in carries. 111 issues in openai/codex quote it as of September 26, 2026, and three causes have been confirmed.
The model was retired from ChatGPT sign-in
This is the June 2026 wave. Plus subscribers using gpt-5.3-codex started getting this line in early June, and an OpenAI maintainer answered in #17642 on June 8: "the 5.2 and 5.3-codex models have been retired. You'll need to switch to a newer model."
The next one has a date. OpenAI's Codex models page says GPT-5.5 leaves ChatGPT, ChatGPT Work and Codex on every plan on October 14, 2026 (the API keeps it). The replacement it names is gpt-6-sol on Plus and above, and gpt-6-luna on Free and Go in the desktop app.
The model needs a higher plan
Some models are gated by plan. When the same error hit gpt-5.3-codex-spark in April, the maintainer replied that the model "is currently supported only for Pro subscriptions" and asked whether the subscription had lapsed back to free. A lapsed subscription or a move from Pro down to Plus lands you here for any Pro-only model.
The model hasn't reached your account
The September reports are the murkiest. In #46304 a Pro subscriber got this line for gpt-5.6-sol and gpt-6-astra in the desktop app, while ChatGPT Work worked on chatgpt.com for the same account, and says OpenAI Support called it a backend access issue. There's nothing to fix on your side in that case, beyond picking a model that works and reporting it.
Clearing the old name out
The model name is sticky, which is why the error can survive a re-login. Codex reads it from /model, from -m or --model at launch, and from the model key in ~/.codex/config.toml (or a profile inside it). One commenter fixed Spark by resetting config.toml and switching models. OpenAI's own retirement note tells workspaces to replace the old name in "saved model settings, managed configurations, custom agents, scheduled tasks, and scripts", since each of those can keep asking for a model that's gone.
If the reply is unexpected status 404 Not Found: The model ... does not exist or you do not have access to it instead, it's the same family of problem with a different code. For a model that exists but is busy, Codex says “Selected model is at capacity”.
Other lines the same feature prints
Match yours against these if the one at the top of the page is not quite it. They come from the same code and mean related things.
The 'gpt-5.3-codex' model is not supported when using Codex with a ChatGPT account.The 'gpt-5.3-codex-spark' model is not supported when using Codex with a ChatGPT account.The 'gpt-6-astra' model is not supported when using Codex with a ChatGPT account.{"type":"error","status":400,"error":{"type":"invalid_request_error","message":"The 'gpt-5.3-codex' model is not supported when using Codex with a ChatGPT account."}}