openclaw - ✅(Solved) Fix 2026.4.26: openai-codex/gpt-5.4-mini listed as available but cron fails with ChatGPT account unsupported [1 pull requests, 1 comments, 2 participants]

Official PRs (…)
ON THIS PAGE

Recommended Tools

×6

Utilities matched from this issue’s tags and category — try them while you read without losing context.

GitHub issue graph ai analysis

Paste a GitHub issue URL. We fetch that issue, discover linked issues from bodies/comments/timeline, collect linked pull requests, and produce a structured English report.

The report is written in English Markdown for sharing and archival.

Helpful · Quick feedback

Loading…
GitHub stats
openclaw/openclaw#73242Fetched 2026-04-29 06:21:56
View on GitHub
Comments
1
Participants
2
Timeline
7
Reactions
0
Author
Participants
Timeline (top)
referenced ×2closed ×1commented ×1cross-referenced ×1

Body:

OpenClaw 2026.4.26 (be8c246)

After updating to .26, isolated cron jobs using `openai-codex/gpt-5.4-mini` started failing.

`openclaw models list --provider openai-codex` shows:

- openai-codex/gpt-5.4
- openai-codex/gpt-5.5
- openai-codex/gpt-5.4-mini

`gpt-5.4-mini` is shown as configured/authenticated, but the cron run fails with:

{"detail":"The 'openai-codex/gpt-5.4-mini' model is not supported when using Codex with a ChatGPT account."}

Example cron job:

- id: 2c72759b-e54a-43a2-bad6-73cbf1deaae3
- name: Polymarket Weather Scan
- session: isolated
- model: openai-codex/gpt-5.4-mini

Expected:
Either the model should work, or `models list` / cron validation should not present it as usable for Codex ChatGPT OAuth.

Actual:
Cron resolves the configured model, starts the run, then fails at runtime with upstream unsupported-model error.

Worked in 23.4.2026, broke in 26.4.2026

Error Message

Cron resolves the configured model, starts the run, then fails at runtime with upstream unsupported-model error. On 26.4.2026 - Try use openai-codex/gpt-5.4-mini and it will throw that error

Root Cause

Body:

OpenClaw 2026.4.26 (be8c246)

After updating to .26, isolated cron jobs using `openai-codex/gpt-5.4-mini` started failing.

`openclaw models list --provider openai-codex` shows:

- openai-codex/gpt-5.4
- openai-codex/gpt-5.5
- openai-codex/gpt-5.4-mini

`gpt-5.4-mini` is shown as configured/authenticated, but the cron run fails with:

{"detail":"The 'openai-codex/gpt-5.4-mini' model is not supported when using Codex with a ChatGPT account."}

Example cron job:

- id: 2c72759b-e54a-43a2-bad6-73cbf1deaae3
- name: Polymarket Weather Scan
- session: isolated
- model: openai-codex/gpt-5.4-mini

Expected:
Either the model should work, or `models list` / cron validation should not present it as usable for Codex ChatGPT OAuth.

Actual:
Cron resolves the configured model, starts the run, then fails at runtime with upstream unsupported-model error.

Worked in 23.4.2026, broke in 26.4.2026

Fix Action

Fixed

PR fix notes

PR #73280: fix(codex): hide ChatGPT-account-unsupported models from catalog (#73242)

Description (problem / solution / changelog)

What

openclaw models list --provider openai-codex and isolated cron model resolution surface openai-codex/gpt-5.4-mini even when the only configured openai-codex auth is a ChatGPT-OAuth (subscription) credential. The upstream Codex app-server then rejects the run with:

The 'openai-codex/gpt-5.4-mini' model is not supported when using Codex with a ChatGPT account.

This regressed in 2026.4.26 and is reported in #73242.

Closes #73242.

Why this fix

augmentModelCatalog in extensions/openai/openai-codex-provider.ts unconditionally adds gpt-5.4-mini (and the other synthetic Codex rows) to the catalog as long as any openai-codex profile exists. Per extensions/CLAUDE.md, vendor-specific catalog selection lives in the owner extension; per the src/plugins/CLAUDE.md boundary rules, this stays plugin-owned and does not require a generic core seam.

The fix introduces a small list of ChatGPT-account-unsupported model ids and filters them from the synthetic catalog when:

  1. ctx.agentDir is provided (otherwise we cannot read the auth store; we keep current behavior),
  2. neither OPENAI_API_KEY nor CODEX_API_KEY is set in the env, and
  3. no openai-codex profile in the auth store has type: "api_key".

If any of those conditions indicate API-key access is available, the model stays visible. This keeps the existing "include gpt-5.4-mini for API-key Codex users" behavior intact while hiding the model from the ChatGPT-OAuth-only path that cannot invoke it.

The "ChatGPT account unsupported" model list is currently ["gpt-5.4-mini"], matching exactly the model named in #73242. New ids can be added later if other rows show the same upstream rejection.

Tests

Added three cases in extensions/openai/openai-codex-provider.test.ts:

  • ChatGPT-OAuth-only profile, no env api key, with agentDir: gpt-5.4-mini is filtered out; gpt-5.4, gpt-5.5-pro, etc. still surface.
  • API-key profile present: gpt-5.4-mini stays.
  • ChatGPT-OAuth-only profile but OPENAI_API_KEY is set: gpt-5.4-mini stays, and the auth store is not even read (env short-circuits).

The existing "augments catalog with gpt-5.5-pro and gpt-5.4 native metadata" test does not pass agentDir, so it continues to assert that gpt-5.4-mini is included — preserving the no-knowledge-of-auth default.

Notes

  • Single-area diff: extensions/openai/openai-codex-provider.ts, its test, and a one-line CHANGELOG entry under 2026.4.27 ### Fixes with a Thanks @edwin-rivera-dev credit.
  • No core, plugin SDK, or other-extension files touched. Owner-boundary respected.
  • AI-assisted (Claude). Reviewed locally; please flag any oversight.

Validation

Local CI is constrained on this machine, so I have not run the full pnpm test/pnpm check:changed lanes. I have:

  • Reviewed the diff against the existing augmentModelCatalog and auth-profile patterns already used at the catalog run site (line 487 of the same file).
  • Verified the new helpers do not change behavior when ctx.agentDir is undefined (the existing test case).
  • Confirmed the CHANGELOG entry is single-line and credits a contributor, per CLAUDE.md.

Happy to run additional Testbox lanes (pnpm test:extension openai, pnpm check:changed) if you can dispatch a green run; the touched surface is just the openai extension and the changelog.

Changed files

  • CHANGELOG.md (modified, +1/-0)
  • extensions/openai/openai-codex-provider.test.ts (modified, +143/-0)
  • extensions/openai/openai-codex-provider.ts (modified, +61/-1)

Code Example

OpenClaw 2026.4.26 (be8c246)

After updating to .26, isolated cron jobs using `openai-codex/gpt-5.4-mini` started failing.

`openclaw models list --provider openai-codex` shows:

- openai-codex/gpt-5.4
- openai-codex/gpt-5.5

---

- openai-codex/gpt-5.4-mini

`gpt-5.4-mini` is shown as configured/authenticated, but the cron run fails with:

{"detail":"The 'openai-codex/gpt-5.4-mini' model is not supported when using Codex with a ChatGPT account."}

Example cron job:

- id: 2c72759b-e54a-43a2-bad6-73cbf1deaae3
- name: Polymarket Weather Scan
- session: isolated
- model: openai-codex/gpt-5.4-mini

Expected:
Either the model should work, or `models list` / cron validation should not present it as usable for Codex ChatGPT OAuth.

---

Actual:
Cron resolves the configured model, starts the run, then fails at runtime with upstream unsupported-model error.

---
RAW_BUFFERClick to expand / collapse

Bug type

Regression (worked before, now fails)

Beta release blocker

No

Summary

Body:

OpenClaw 2026.4.26 (be8c246)

After updating to .26, isolated cron jobs using `openai-codex/gpt-5.4-mini` started failing.

`openclaw models list --provider openai-codex` shows:

- openai-codex/gpt-5.4
- openai-codex/gpt-5.5
- openai-codex/gpt-5.4-mini

`gpt-5.4-mini` is shown as configured/authenticated, but the cron run fails with:

{"detail":"The 'openai-codex/gpt-5.4-mini' model is not supported when using Codex with a ChatGPT account."}

Example cron job:

- id: 2c72759b-e54a-43a2-bad6-73cbf1deaae3
- name: Polymarket Weather Scan
- session: isolated
- model: openai-codex/gpt-5.4-mini

Expected:
Either the model should work, or `models list` / cron validation should not present it as usable for Codex ChatGPT OAuth.

Actual:
Cron resolves the configured model, starts the run, then fails at runtime with upstream unsupported-model error.

Worked in 23.4.2026, broke in 26.4.2026

Steps to reproduce

On 26.4.2026 - Try use openai-codex/gpt-5.4-mini and it will throw that error

Expected behavior

Read above

Actual behavior

Read above

OpenClaw version

2026.4.26

Operating system

Ubuntu

Install method

No response

Model

Openai-codex/gpt-5.4-mini

Provider / routing chain

Openclaw

Additional provider/model setup details

No response

Logs, screenshots, and evidence

Impact and severity

No response

Additional information

No response

extent analysis

TL;DR

The openai-codex/gpt-5.4-mini model is not supported with Codex ChatGPT accounts, causing cron jobs to fail after updating to OpenClaw 2026.4.26.

Guidance

  • Verify that the openai-codex/gpt-5.4-mini model is compatible with the current Codex account type, as the error message suggests it's not supported.
  • Check the OpenClaw documentation for any updates on supported models and account types.
  • Consider using a different model, such as openai-codex/gpt-5.4 or openai-codex/gpt-5.5, which are listed as available in the openclaw models list output.
  • Review the cron job configuration to ensure it's correctly set up for the intended model and account type.

Notes

The issue seems to be related to a change in supported models between OpenClaw versions 23.4.2026 and 26.4.2026. Without further information on the Codex account type and setup, it's difficult to provide a more specific solution.

Recommendation

Apply workaround: Use a different model, such as openai-codex/gpt-5.4 or openai-codex/gpt-5.5, which are listed as available and may be compatible with the current Codex account type.

Vote matrix · Quick signals

Works
Did the solution work? Tap to confirm.
Easy Fix
Was it a quick fix?
Time Saver
Did it save you time?
Blocking
Was it severely blocking?
Common Issue
Are others likely hitting this too?
Flaky / Intermittent
Is it intermittent?
Verified / Reproducible
Can you reproduce it reliably?
Loading…

FAQ

Expected behavior

Read above

Still need to ship something?

×6

Another batch ranked right after the header list — different links, same matching logic.

Back to top recommendations

TRENDING