Skip to content
TheChatArchiveThe Chat Archive
gpt-5.6-sol· openaiHugging 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.

    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. Assistant

    I’ll inspect the repository structure and the version-resolution/model-refresh code first, then identify relevant tests and reproduce the fallback behavior before choosing the least invasive robust fix.

  3. Assistant

    The current implementation only tries GitHub and keeps cache in memory, exactly matching the report. I’ll inspect its unit tests, networking patterns, and package configuration; a registry.npmjs.org fallback is likely the most targeted fix because it removes dependence on GitHub’s anonymous rate limit without requiring persistence/schema changes.

  4. Assistant

    Existing tests confirm the intended cache semantics but do not cover a secondary source. Before editing, I’ll reproduce the reported condition with a temporary inline script: GitHub returns 403, and despite an available npm response the current code performs only one request and returns 0.101.0. Then I’ll refactor source fetching so failures and malformed responses are isolated per source while preserving stale-cache and settings fallback behavior.

  5. Assistant

    The reproduction returned the stale default and made only one HTTP request. I’ll now add the npm package metadata endpoint as the second source. Each request will have independent exception/status/schema handling, so a GitHub 403, invalid GitHub payload, or GitHub network error can still proceed to npm; only failure of both sources reaches the existing stale/default fallbacks.

  6. Assistant

    The implementation now preserves the fast GitHub-success path and only contacts npm when needed. I’ll run the focused unit suite plus the same reproduction harness to verify both backward compatibility and that a GitHub 403 now resolves to the npm version. I’ll also lint/type-check the changed file afterward.

  7. Assistant

    Focused tests and the exact 403 reproduction pass. I’ll verify the chosen npm URL and response shape against the live registry, then run the broader unit test suite because this cache feeds model refresh globally. I’ll also inspect model-fetch call sites to ensure no assumptions were changed by returning the secondary version.

  8. Assistant

    The live endpoint returns the expected package/version (0.144.5). The broad unit run reached 1,887 passes but one unrelated database test failed under high-parallel xdist because its worker had no accounts table; the changed module is not involved. I’ll rerun that test in isolation to distinguish an xdist fixture race from a real regression, and run static type checking on the changed module. I’ll also inspect the final diff for unnecessary behavioral changes.

  9. Assistant

    The isolated unrelated test fails identically, confirming a pre-existing fixture/schema issue rather than this change. One refinement is warranted: because the outer try currently covers session cleanup too, a cleanup exception after a successful source could discard the resolved version. I’ll simplify session lifecycle handling only if needed after checking project Python/type conventions; then run model-fetcher/model-registry tests and a final focused suite.

  10. Assistant

    The codebase uses ty rather than mypy. The lifecycle edge case is theoretical and the current broad guard safely handles both construction and cleanup without leaking exceptions, so I’ll avoid needless restructuring. I’ll run ty on the file and all directly affected model-fetch/registry tests, then review the exact final patch.

  11. Assistant

    Final verification is clean. Checklist review before submission:

    • Reproduced the reported error: yes—the pre-change harness simulated GitHub HTTP 403 with npm metadata available; it resolved 0.101.0 and made only the GitHub request.
    • Confirmed the fix resolves it: yes—the same harness now makes GitHub then npm requests and resolves 0.130.0; the live npm endpoint also returns valid current metadata.
    • Considered edge cases: yes—successful GitHub remains one request; per-source HTTP errors, network/JSON exceptions, malformed/non-stable versions, both-source failure, stale-cache reuse, configured fallback, TTL, invalidation, proxy environment handling, and concurrent locking retain safe behavior. Existing focused tests exercise these fallback/cache paths.
    • Ran affected existing tests: yes—34 directly affected version/model-fetch/refresh/registry tests pass; Ruff formatting/lint, ty, and diff checks pass. The broader unit run had 1,887 passes and one unrelated request-log database test failure (no such table: accounts), which reproduces in isolation and does not import or exercise this change.

    I’ll submit the single-file implementation now.

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