Vision capabilities are not detected properly for qwen/qwen3-vl-4b

[tholu]

Using the vision model qwen/qwen3-vl-4b with LM Studio but it doesn't recognize the vision capability:

http://localhost:40114/olla/models

    {
      "olla": {
        "family": "qwen",
        "variant": "3",
        "parameter_size": "",
        "quantization": "",
        "aliases": [
          "qwen/qwen3-vl-4b"
        ],
        "availability": [
          {
            "endpoint": "lm-studio",
            "state": "unknown"
          },
        ],
        "capabilities": [
          "text-generation"
        ]
      },
      "id": "qwen/qwen3-vl-4b",
      "object": "model",
      "owned_by": "olla",
      "created": 1772147744
    }

[thushan]

hey,

thanks for the issue, we are working on updating the model capabilities for 2026. It's a bit behind but I'll see if we can fix this soon.

There's a bit of rework in the capability detection being done, which is why its taking sometime. Maybe a good usecase to test when we have a test build - if you're able?

[tholu]

@thushan Of course, happy to test!