Skip to content

[FEATURE]:opencode #64536

Closed as duplicate
Closed as duplicate

Description

@deepseektest364-alt

Feature hasn't been suggested before.

  • I have verified this feature I'm about to request hasn't been suggested before.

Describe the enhancement you want to request

Hello! I want the following:

NVIDIA provider has added new model. Add Mistral model in Provider: NVIDIA

"model": "mistralai/mistral-large-4-1t-a52b"

Alongside that, certain models do not work and are deprecated, or other reasons.

Nvidia/nemotron-voicechat gives answer "Provider request failed with HTTP 404: 404 page not found"
Nvidia/paligemma gives answer "Provider request failed with HTTP 404: 404 page not found"
Nvidia/Llama Guard 4 12 B gives answer "The model 'meta/llama-guard-4-12b' has reached its end of life on 2026-11-16T09:00:00Z and is no longer available."

Alongside that, following models, say the same too:
The model 'google/diffusiongemma-26b-a4b-it' has reached its end of life on 2026-11-12T09:00:00Z and is no longer available.

Activity

  1. opencodejr commented on Nov 17, 2026

    @opencodejr
    Contributor

    Thanks for the details, @deepseektest364-alt! I think this is a duplicate of #40524. That issue asks OpenCode to check each provider's own /models endpoint, so new models (like mistralai/mistral-large-4-1t-a52b) become selectable and models the provider no longer serves (like the 404 and end-of-life ones you list) stop being offered.

    Related issues, ordered by likelihood:

    The built-in model list comes from models.dev. Until it catches up, you can add a missing model and hide retired ones yourself in your config's models map (providers docs):

    {
      "providers": {
        "nvidia": {
          "models": {
            "mistralai/mistral-large-4-1t-a52b": { "name": "Mistral Large 4" },
            "meta/llama-guard-4-12b": { "disabled": true }
          }
        }
      }
    }

    The keys must match the model IDs in your model list. If a key doesn't match an existing entry, it adds a new model instead of hiding that one.

    I'm closing this issue now as a duplicate of #40524. Please add your details there (the new model ID and the retired ones). If you think this is a different problem, reply here and the issue will reopen.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions