Skip to content
TheChatArchiveThe Chat Archive
gpt-5.6-sol· openaiHugging Face ibragim-bad/swe_rebench_07_2026_trajectories

/v1/models: return the full catalog with per-model availability + opt-in filter Split…

  1. User

    /v1/models: return the full catalog with per-model availability + opt-in filter Split out from #242.

    Current: `GET /v1/models` returns only models that are connected right now (enabled AND an enabled key can serve them).

    Wanted: by default return the whole catalog (one entry per model id), each annotated with whether it's usable, plus an opt-in filter to get just the connected ones.

    Per-model annotation:

    • `available`: boolean — enabled AND a matching enabled key exists.
    • `unavailable_reason`: `null` | `"no_key"` (enabled, no key) | `"disabled"` (catalog row disabled).

    Filter:

    • `?available=true` (alias `?connected=true`) → only connected models (the previous default behavior).

    Notes:

    • The extra fields are additive — OpenAI-compatible clients ignore unknown fields.
    • Dedup prefers an available instance of a model id over a disabled/keyless one.
    • The `auto` virtual model is always `available: true`.

    Interface

    No new interfaces are introduced.

  2. Assistant

    I’ll trace the models endpoint and its tests, then update the implementation and verify both default catalog and opt-in filtering behavior.

  3. Assistant

    rg isn’t installed in this workspace, so I’m falling back to targeted find/grep inspection.

  4. Assistant

    The endpoint currently filters in SQL before deduplication. I’m changing that to annotate every catalog row first, deduplicate by model ID with availability precedence, and apply available=true/connected=true only after deduplication. I’ll also make the virtual auto entry explicitly always available.

  5. Assistant

    The focused endpoint tests now pass, including unavailable reasons, both filter names, and availability-first deduplication. I’m running the TypeScript build and full server suite next to catch integration regressions.

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