openai / openai/codex

gpt-5.5` returns 404 on the Codex backend route while the same account serves it fine in the ChatGPT app

Open
#44,477 4 comments 1 reaction 0 assignees View on GitHub

Nobody has claimed this yet.

app bug CLI connectivity exec
Dominant language
Rust
Stars
125k
Forks
19.4k
PR merge metrics
PR metrics pending

Description

What version of the Codex App are you using (From “About Codex” dialog)?

0.153.4

What subscription do you have?

max $200

What platform is your computer?

macOS

What issue are you seeing?

gpt-5.5 returns 404 on the Codex backend route while the same account serves it fine in the ChatGPT app

Summary

Since 2026-09-06 ~23:10 EDT, codex exec --model gpt-5.5 intermittently (now almost
continuously) fails with:

ERROR: unexpected status 404 Not Found: The model `gpt-5.5` does not exist or you do not
have access to it., url: https://chatgpt.com/backend-api/codex/responses

The same ChatGPT account can still use GPT-5.5 normally in the ChatGPT app at the same
time, and other models (gpt-5.6-sol, gpt-5.6-terra, gpt-5.6-luna) work fine through
Codex on the identical setup. This looks like a model-registry/routing problem specific to
the Codex backend route (backend-api/codex/responses), similar in signature to the
June 2026 reports (#26892, #26910, #26916, #29546).

Environment

  • codex-cli 0.153.4 (also reproduced on 0.151.0-alpha.7.2 before updating — updating did not help)
  • Auth: ChatGPT account (not API key)
  • macOS (Darwin 25.5.0), binary from ChatGPT.app (/Applications/ChatGPT.app/Contents/Resources/codex)
  • Region: cf-ray suffix DTW (Detroit)
  • Docs still list gpt-5.5 as available for Codex (previous-generation flagship), no retirement date

Repro

echo "" | codex exec --skip-git-repo-check --model gpt-5.5 -c model_reasoning_effort=low "Reply OK"
# -> 404 "The model `gpt-5.5` does not exist or you do not have access to it."

echo "" | codex exec --skip-git-repo-check --model gpt-5.6-sol -c model_reasoning_effort=low "Reply OK"
# -> OK   (same machine, same account, same minute)

Evidence / request IDs

when (EDT) cf-ray request id
2026-09-09 15:59 a388bf61f97c1d91-DTW 26901686-8c36-450a-ab71-f3ce446e493e
2026-09-10 05:12 a38d4f3bcf852bdb-DTW 237d4b81-b771-4997-a018-1ebbedab7dd6

Availability pattern (from an automated 5-minute prober running since 09-06)

The model is not fully down — it serves in short windows, then 404s again:

  • 09-06 23:12 EDT: first 404 (mid-run; had worked ~1h earlier)
  • 09-07 → 09-09: mostly 404 with occasional serving windows (e.g. 09-09 10:49, 16:09–16:24, brief windows 16:35, 16:54, 17:06)
  • 09-10 00:40 EDT: brief serving window (minutes), then 404 again through the night
  • 09-10 05:12 EDT: still 404

The on/off pattern plus the app-vs-Codex asymmetry suggests a propagation/registry issue on
the Codex serving route rather than a model retirement or account entitlement change.

Impact

Long-running scientific extraction pipelines pinned to gpt-5.5 (for reproducibility across a
multi-week study) have been blocked or crawling through retry windows for 4 days. Switching
models invalidates the study's validation baseline, so a fix or an official statement on
gpt-5.5's Codex availability would be very helpful.

What steps can reproduce the bug?

gpt-5.5 returns 404 on the Codex backend route while the same account serves it fine in the ChatGPT app

Summary

Since 2026-09-06 ~23:10 EDT, codex exec --model gpt-5.5 intermittently (now almost
continuously) fails with:

ERROR: unexpected status 404 Not Found: The model `gpt-5.5` does not exist or you do not
have access to it., url: https://chatgpt.com/backend-api/codex/responses

The same ChatGPT account can still use GPT-5.5 normally in the ChatGPT app at the same
time, and other models (gpt-5.6-sol, gpt-5.6-terra, gpt-5.6-luna) work fine through
Codex on the identical setup. This looks like a model-registry/routing problem specific to
the Codex backend route (backend-api/codex/responses), similar in signature to the
June 2026 reports (#26892, #26910, #26916, #29546).

Environment

  • codex-cli 0.153.4 (also reproduced on 0.151.0-alpha.7.2 before updating — updating did not help)
  • Auth: ChatGPT account (not API key)
  • macOS (Darwin 25.5.0), binary from ChatGPT.app (/Applications/ChatGPT.app/Contents/Resources/codex)
  • Region: cf-ray suffix DTW (Detroit)
  • Docs still list gpt-5.5 as available for Codex (previous-generation flagship), no retirement date

Repro

echo "" | codex exec --skip-git-repo-check --model gpt-5.5 -c model_reasoning_effort=low "Reply OK"
# -> 404 "The model `gpt-5.5` does not exist or you do not have access to it."

echo "" | codex exec --skip-git-repo-check --model gpt-5.6-sol -c model_reasoning_effort=low "Reply OK"
# -> OK   (same machine, same account, same minute)

Evidence / request IDs

when (EDT) cf-ray request id
2026-09-09 15:59 a388bf61f97c1d91-DTW 26901686-8c36-450a-ab71-f3ce446e493e
2026-09-10 05:12 a38d4f3bcf852bdb-DTW 237d4b81-b771-4997-a018-1ebbedab7dd6

Availability pattern (from an automated 5-minute prober running since 09-06)

The model is not fully down — it serves in short windows, then 404s again:

  • 09-06 23:12 EDT: first 404 (mid-run; had worked ~1h earlier)
  • 09-07 → 09-09: mostly 404 with occasional serving windows (e.g. 09-09 10:49, 16:09–16:24, brief windows 16:35, 16:54, 17:06)
  • 09-10 00:40 EDT: brief serving window (minutes), then 404 again through the night
  • 09-10 05:12 EDT: still 404

The on/off pattern plus the app-vs-Codex asymmetry suggests a propagation/registry issue on
the Codex serving route rather than a model retirement or account entitlement change.

Impact

Long-running scientific extraction pipelines pinned to gpt-5.5 (for reproducibility across a
multi-week study) have been blocked or crawling through retry windows for 4 days. Switching
models invalidates the study's validation baseline, so a fix or an official statement on
gpt-5.5's Codex availability would be very helpful.

What is the expected behavior?

No response

Additional information

No response

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by running the provided codex exec commands with gpt-5.5 and gpt-5.6-sol, recording the 404 response and request identifiers against the backend-api/codex/responses route. Compare the intermittent behavior with the reported ChatGPT app availability; done requires restoring gpt-5.5 on the Codex route or documenting an official availability decision.

Written by the indexing model from the issue text.

Assessment

Tech stack
rust
Domain
api, backend, cli
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Active
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.