The requested model 'gpt-5.3-codex' does not exist

Potential duplicates detected. Please review them and close your issue if it is a duplicate.

• #10781
• #8464

Powered by Codex Action

Should be working fine. I'm not able to repro the problem you're seeing. Do you perhaps have an LLM proxy server configured? Do you have any other non-default settings in your config.toml? Do you have OPENAI_BASE_URL defined?

Just a sec. So what I can say is, that the API Platform is also not yet showing the gpt-5.3-codex anywhere where models are listed. E.g.

• https://platform.openai.com/settings/organization/limits > "Show all models" (button)
• https://platform.openai.com/settings/<project>/limits > Allow or block models (section) > "Select models" (button)

When listing /model in codex I only see

1. gpt-5.2-codex (current)  Latest frontier agentic coding model.
2. gpt-5.2                  Latest frontier model with improvements across knowledge, reasoning and coding
3. gpt-5.1-codex-max        Codex-optimized flagship for deep and fast reasoning.
4. gpt-5.1-codex-mini       Optimized for codex. Cheaper, faster, but less capable.

And my only /cygdrive/c/Users/<user>/.codex/config.toml (equivalent to %userprofile%\.codex\config.toml) is just having

personality = "pragmatic"
model = "gpt-5.2-codex"
model_reasoning_effort = "high"
[features]
powershell_utf8 = true
steer = true

Regarding your questions:

• never heard of OPENAI_BASE_URL, so probably no.
• I use plain vanilla codex, no proxy, nothing similiar

Yes, gpt-5.3-codex is not yet available for API Key usage. You need to use a ChatGPT subscription to access it.

Yes, gpt-5.3-codex is not yet available for API Key usage. You need to use a ChatGPT subscription to access it.

I'm having the same issue when accessing via /model. I'm on a business subscription. See attached.

I was able to manually set it within config.toml file.

<img width="664" height="351" alt="Image" src="https://github.com/user-attachments/assets/cc35e509-bf04-4a2a-a2b3-08bd79ce34c9" />

@Adealia, try rm ~/.codex/models_cache.json and then relaunching. Then use the app for a little while. The models cache should repopulate after a short time.

I found that even though my codex CLI was updated, you might have to update in the app-- which makes it go away (they're separate binaries and codex app seems not to auto-update eagerly)

@vincentzed, yes, the app ships with its own copy of the CLI binary, and that binary's version might be different than the CLI that you installed separately.

Only thing that worked for me was logging out and logging back in. Clearing cache did not help, using other apps/GUI also did not help.

Even loggin /out didnt work out for me ? am i missing something

upgrading to the latest cli version .98 worked

Update Codex to the latest version

npm install -g @openai/codex@latest

@Adealia, try rm ~/.codex/models_cache.json and then relaunching. Then use the app for a little while. The models cache should repopulate after a short time.

the file does not exist here

latest version is installed but no gpt-5.3-codex available...

gpt-5.3-codex is not working as mcp . I deleted /.codex/models_cache.json , restarted vscode, installed latest codex agent version through brew etc. It seems a half backed release. Back to Opus ...

Yes, gpt-5.3-codex is not yet available for API Key usage. You need to use a ChatGPT subscription to access it.

And then you get:
"{"detail":"The '' model is not supported when using Codex with a ChatGPT account."}

this was the message that crashed terminal and exited CLI. This error was displayed in response to question "What model are you using?".

This issue thread is confusing a bunch of different issues.

To access the gpt-5.3-codex model, you need to be using the latest version of the CLI and log in using your ChatGPT account, not your API key.

If you are seeing other issues, please open a new bug report. I'm going to lock this issue to avoid further confusion.