Skip to content
TheChatArchiveThe Chat Archive
Claudeclaude-sonnet-5· anthropicHugging Face ibragim-bad/swe_rebench_07_2026_trajectories

ISSUE DESCRIPTION Stale 'clientversion' fallback causes 503 'noplansupportformodel' for…

  1. User

    ISSUE DESCRIPTION

    Stale 'client_version' fallback causes 503 'no_plan_support_for_model' for newer models (e.g. gpt-5.5) when GitHub releases API rate-limits

    Summary

    When app/core/clients/codex_version.py cannot fetch the latest codex release from api.github.com/repos/openai/codex/releases/latest (e.g. anonymous-rate-limited with HTTP 403), it logs:

    GitHub releases API returned HTTP 403
    GitHub fetch failed and no cached version; falling back to settings default 0.101.0
    

    and falls back to Settings.model_registry_client_version (default 0.101.0). The model registry refresh (app/core/openai/model_refresh_scheduler.py → clients/model_fetcher.fetch_models_for_plan) then asks upstream /codex/models?client_version=0.101.0. Upstream gates newer model slugs behind newer client versions, so the response excludes e.g. gpt-5.5.

    As a result, ModelRegistry.model_plans never gets a gpt-5.5 entry, _filter_accounts_for_model in modules/proxy/load_balancer.py:1168 returns an empty set, and every Codex request for gpt-5.5 is rejected with:

    proxy_error_response status=503 code=no_plan_support_for_model
      message=No accounts with a plan supporting model 'gpt-5.5'
    

    …even though the actual Codex CLI is on 0.130.0 and the account's ChatGPT subscription does support the model (calling /backend-api/codex/models?client_version=0.130.0 through the LB returns 200 with the model present).

    Reproduction

    1. Run ghcr.io/soju06/codex-lb:1.17.0 on a host where anonymous GitHub API requests are 403'ed.
    2. Connect a current Codex CLI (e.g. @openai/codex@0.130.0) configured with model = \"gpt-5.5\".
    3. Observe startup log line:
      • GitHub fetch failed and no cached version; falling back to settings default 0.101.0
    4. Every POST /backend-api/codex/responses returns 503 with code no_plan_support_for_model.

    Setting CODEX_LB_MODEL_REGISTRY_CLIENT_VERSION=0.130.0 and recreating the container fully resolves the symptom.

    Suggested fixes

    A few low-effort options, any one would help:

    1. Bump the baked-in default model_registry_client_version on each release so it tracks the recent @openai/codex version (or read it from a generated file at build time).
    2. Add a secondary version source. npm view @openai/codex version against https://registry.npmjs.org is not rate-limited the way the GitHub anonymous API is, and a static fetch of https://registry.npmjs.org/@openai/codex/latest returns JSON containing \"version\".
    3. Cache the fetched version on disk (e.g. inside the data volume) so a single successful lookup survives container restarts and survives transient 403s for the cache TTL.
    4. Honor the client's own client_version query string as a hint for which model list to refresh (only when greater than the configured default), so a newer client effectively bootstraps the registry.

    Workaround

    For anyone hitting this, pass -e CODEX_LB_MODEL_REGISTRY_CLIENT_VERSION=<your-codex-cli-version> to docker run:

    docker run -d \\
      -e CODEX_LB_DASHBOARD_BOOTSTRAP_TOKEN=... \\
      -e CODEX_LB_MODEL_REGISTRY_CLIENT_VERSION=0.130.0 \\
      -p 2455:2455 -p 1455:1455 \\
      -v codex-lb-data:/var/lib/codex-lb \\
      ghcr.io/soju06/codex-lb:1.17.0
    

    Environment

    • codex-lb image: ghcr.io/soju06/codex-lb:1.17.0
    • Codex CLI: @openai/codex@0.130.0
    • Host: macOS, Docker Desktop / Colima
    • Account plan: pro (3 accounts, all plan_type=pro in store.db)

    Related: #480 ("Model metadata for gpt-5.5 not found. Defaulting to fallback metadata") looks like the same underlying cause.

    Interface

    No new interfaces are introduced.

    ADDITIONAL ADVICE

    Since you are given a git repository, you can use git commands to simplify your work. For example, if you made a mistake and want to revert your changes, you can use git checkout <file> to restore the file to its original state. You can also reset all changes in the repository using git reset --hard command. Additionally, you can use git stash and git stash pop to temporarily save your changes and restore them later. Finally, you can see the changes that you've made (which will be submitted when you call submit) using git status or git diff commands. However you don't need to use git add or git commit before submitting your changes. In fact, submit only submits currently unstaged changes because it uses git diff to get the changes that need to be submitted.

    CHECKLIST

    Before submitting your solution, please go over the following checklist and make sure you've done everything:

    • If an error was reported in the issue description, I have successfully reproduced it.
    • If an error was reported in the issue description, I have confirmed that my fix resolves the error.
    • I have thought about edge cases and made sure my fix handles them as well.
    • I have run existing tests in the repository that might have been affected by the change I made and confirmed that they pass. I want you to list every bullet from this checklist and write your reasoning for why you think you did it or didn't need to.

    Repository has been uploaded and your shell is currently at the repository root. Time to solve the issue!

    (Current directory: /codex-lb, current file: none) bash-$

  2. Claude Assistant

    Now let's add new settings fields for the npm fallback and disk cache path.

  3. Claude Assistant

    Now let's check the field_validator area for encryption_key_file to see if similar Path validators exist, in case we need one for the new path field too.

  4. Claude Assistant

    Now let's run the existing test suite for codex_version to see if it passes, and fix any test expectations that may have changed due to new npm fallback (since existing tests only mock github session, the npm fallback will also try to fetch and might fail due to mocking only one session).

  5. Claude Assistant

    Now remove the disk cache file setting since it's unused, but keep the npm settings. Let's update settings.py.

  6. Claude Assistant

    Now let's run the codex_version tests again.

  7. Claude Assistant

    All 11 pass now. Let's also run broader test suite affected areas (settings tests, model_fetcher, model_refresh_scheduler).

  8. Claude Assistant

    Let's also check for any settings tests specifically to make sure new fields don't break anything.

  9. Claude Assistant

    Now let's write a quick reproduction script to simulate the reported scenario: GitHub 403, npm success.

  10. Claude Assistant

    This confirms the fix: when GitHub returns 403, the npm registry fallback correctly resolves the current version instead of falling back to the stale settings default. Let's clean up the temp script and also run the full unit test suite to be safe.

  11. Claude Assistant

    All 1889 unit tests pass (unrelated skips/warnings are pre-existing). Now let's also check docs for env var documentation to update (README/docs) mentioning model_registry_client_version, so we can document the new npm fallback settings.

  12. Claude Assistant

    It's a partial list (42 of many settings), and model_registry_client_version isn't in there either, so no update needed there. Let's view the final diff of our changes.

  13. Claude Assistant

    That looks correctly formatted actually (two blank lines is PEP8 standard for top-level definitions). Let's check with ruff/black if available.

  14. Claude Assistant

    Both files pass lint/format checks. Now let's also check mypy/type checking if configured.

  15. Claude Assistant

    Let's run ty check on the modified files if available.

  16. Claude Assistant

    Excellent. Now let's do a final full review of the modified files.

  17. Claude Assistant

    This looks good. Now let's also check get_codex_version_cache usage across the codebase to see whether other components (like model_fetcher) need adjustments — probably not since only version resolution logic changed, same interface.

  18. Claude Assistant

    Good, interface unchanged. Let's run the full test suite one more time to be thorough, plus check test_settings_cache.py more specifically for any settings snapshot tests that enumerate all fields (which might need updates but we shouldn't need to touch them).

  19. Claude Assistant

    No references there. Let's run the complete unit test suite once more to confirm everything is fine after all edits.

  20. Claude Assistant

    All 1889 pass, unrelated skips/warnings pre-existing. Let's review the checklist now.

  21. Claude Assistant

    Everything checks out. Let me review the checklist one final time explicitly before submitting.

    Checklist review:

    • Reproduced the error: Yes — I wrote a temporary repro script mocking a GitHub 403 response (matching the issue's exact scenario) and confirmed that with the old code, this would fall through to the stale 0.101.0 settings default (verified via the existing test test_fallback_to_settings_default_when_no_cache, which encodes this exact behavior).
    • Confirmed fix resolves the error: Yes — the same repro script, run against the patched code, now resolves to 0.130.0 via the npm registry secondary source instead of falling back to the stale default, mirroring the real-world fix described in the issue (CODEX_LB_MODEL_REGISTRY_CLIENT_VERSION=0.130.0 workaround becomes unnecessary since the npm registry isn't rate-limited like GitHub's anonymous API).
    • Edge cases considered: npm response malformed/missing version key, non-200 status from npm, network exceptions during npm fetch, invalid version format from npm (reusing the same _VERSION_RE validation as GitHub), npm fallback disabled via new settings flag (model_registry_version_npm_fallback_enabled, default True but overridable), npm endpoint URL configurable (model_registry_version_npm_registry_url) for testing/mirroring, and preserved existing precedence (GitHub primary → npm secondary → stale in-memory cache → settings default) so behavior only improves, never regresses.
    • Ran existing tests: Yes — ran tests/unit/test_codex_version.py, tests/unit/test_model_fetcher.py, tests/unit/test_model_refresh_scheduler.py, all settings tests, and the full tests/unit suite (1889 passed, 39 pre-existing skips), plus ruff check, ruff format --check, and ty check on the modified files — all clean.

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.