"OpenAI rejected the request"

Running into this too, and have access to all models.

Ran into this exact issue

Can confirm

++ Same issue

Surely there must be a step somewhere where I can provide my API key? The README.md just says to run the npm command and there's "zero configuration" from there. How could that be? Is OpenAI providing this service for free?

In the OP's description they mention OPENAI_API_KEY because it's only in the Non‑interactive / CI mode section. Should we have that in our environment anyway?

I have that but still get this result.

same

same

as well

Happening here as well, maybe still rolling out? I'm on tier 5 and have access via the API and playground.

same pinch.

samesies! <3 super pumepd though!

it prob cant find the key because workdir is defaulted to "~" anyone see how to specify workdir? │ localhost session: * │ ↳ workdir: ~ │ ↳ model: o4-mini │ ↳ approval: suggest

wrong issue, but just run codex on the workdir you want, ex:
cd code/react/foo; codex

This is the issue actually:

    system
    ⚠️  OpenAI rejected the request (request ID: req_a057493e36bb3dda8a2fe4646b7d4cb0): 400 Your organization must be verified to generate reasoning summaries. Please go to: 
https://platform.openai.com/settings/organization/general and click on Verify Organization.. Please verify your settings and try again.

https://platform.openai.com/settings/organization/general

I made a PR to fix this. https://github.com/openai/codex/pull/23

Same problem here!

Hey! 👋

So just a thought — the default model used by the CLI is _o4-mini_, which I think might not be fully rolled out to everyone yet. If your API key doesn’t have access to that model, that could explain the rejection error you’re seeing.

You might want to try running the /model command in the CLI and picking something like _gpt-4.5_ instead — basically any model you know you have access to — and see if that helps.

Not totally sure that’s the issue, but worth trying! Let me know if that works for you.

Hey! 👋 So just a thought — the default model used by the CLI is _o4-mini_, which I think might not be fully rolled out to everyone yet. If your API key doesn’t have access to that model, that could explain the rejection error you’re seeing. You might want to try running the /model command in the CLI and picking something like _gpt-4.5_ instead — basically any model you know you have access to — and see if that helps. Not totally sure that’s the issue, but worth trying! Let me know if that works for you.

I tried with o3-mini and it has the same issue.

Switched to 4o and it fixed!

Anyone else facing the same issue with a different OPENAI_BASE_URL ? I've got a LiteLLM server with O3-mini connected to it, but face the same issue

Hey! 👋 So just a thought — the default model used by the CLI is _o4-mini_, which I think might not be fully rolled out to everyone yet. If your API key doesn’t have access to that model, that could explain the rejection error you’re seeing. You might want to try running the /model command in the CLI and picking something like _gpt-4.5_ instead — basically any model you know you have access to — and see if that helps. Not totally sure that’s the issue, but worth trying! Let me know if that works for you.

this is indeed the problem

o4-mini-2025-04-16 works! and it works ... amazing this is awesome !

It seems to be a rollout issue.

o3 nor o3-2025-04-16 work though.

gpt-4.1 and o4-mini-2025-04-16 seem to work.

EDIT:
After verifying my organization, o3 works

> Hey! 👋 > So just a thought — the default model used by the CLI is _o4-mini_, which I think might not be fully rolled out to everyone yet. If your API key doesn’t have access to that model, that could explain the rejection error you’re seeing. > You might want to try running the /model command in the CLI and picking something like _gpt-4.5_ instead — basically any model you know you have access to — and see if that helps. > Not totally sure that’s the issue, but worth trying! Let me know if that works for you. this is indeed the problem o4-mini-2025-04-16 works! and it works ... amazing this is awesome !

how do you change the model?

> > Hey! 👋 > > So just a thought — the default model used by the CLI is _o4-mini_, which I think might not be fully rolled out to everyone yet. If your API key doesn’t have access to that model, that could explain the rejection error you’re seeing. > > You might want to try running the /model command in the CLI and picking something like _gpt-4.5_ instead — basically any model you know you have access to — and see if that helps. > > Not totally sure that’s the issue, but worth trying! Let me know if that works for you. > > > this is indeed the problem > o4-mini-2025-04-16 works! and it works ... amazing this is awesome ! how do you change the model?

/model

Happens here same issue

Happens here same issue

seea above for solution

@MadcowD is right: the issue is that this is trying to generate reasoning summaries, and you need to verify your org before you can do that.

Right now, non-reasoning models work, but reasoning models will not. If you verify your org (https://platform.openai.com/settings/organization/general), the reasoning models will work.

I can confirm it works, just start codex with

codex -m gpt-4o

I can confirm it works, just start codex with `` codex -m gpt-4o ``

better to use codex -m o4-mini-2025-04-16 to use new 04 mini

verifying our org helped!

https://help.openai.com/en/articles/10910291-api-organization-verification

from blog post
!Image

Can confirm o4-mini-2025-04-16 works.

It does not work, I don't know if it is because I am not in USA.

!Image

Verify your organization people, that is the fix.

Below worked for me.

codex -m o4-mini-2025-04-16

To expand on @ReyNeil's suggestion to "verify your organization," go to https://platform.openai.com/settings/organization/general

There you will see a "Verify Organization" button if you have not verified already.

You may also need to generate and use a new API key (you shouldn't have to but empirically...)

o4-mini should work and for o3 you currently need to verify as explained just above

FWIW I needed to verify, then had the same issue until I generated a new API key post-verification.

Yeah no thanks. Not providing my biometrics, going back to claude code.

!Image

same for me, only o4-mini works and o3 is rejected.

anyone figure out how to get reasoning level to work?

Fix was fixed earlier today, please open new issues for other things and thanks for being patient!

same only the o4-mini-[date-slug] works for me!