Hugging Face Model Not Supported

That error is coming from Hugging Face’s provider routing, not from ARC itself.

What it means

Qwen/Qwen2.5-VL-7B-Instruct is a vision-language model, and Hugging Face is saying:

"You don’t currently have any enabled provider that can serve this model."

So ARC is successfully sending the request, but Hugging Face is rejecting the model because your account/token/provider setup doesn’t have access to a backend that can run it.

What to check

1) Verify the model supports the task you chose

In the Hugging Face skill configuration, make sure you’re using the correct task:

• Vision Chat for asking the model to see and respond
• Object Detection for finding faces/objects in the image
• Image Classification for labeling the whole image
• Image Captioning for describing the scene

If you enabled face and motion tracking, that sounds more like Object Detection, not Vision Chat.

2) Try a known-good default model for the selected task

In the skill’s kebab/gear menu configuration, switch to Default for that task and test again.

For example:

• Object Detection: facebook/detr-resnet-50
• Image Captioning: Salesforce/blip-image-captioning-large

These are generally easier to get working than a large vision-chat model.

3) Check Hugging Face model access / provider availability

That specific error usually means one of these:

• Your token is valid, but the model is not available on the enabled provider
• The model may require a different provider than the one currently enabled in your Hugging Face account
• You may need to select a different model that is supported by serverless inference

Hugging Face has a support article for model-based robot use in ARC here:

• https://synthiam.com/Support/Skills/Artificial-Intelligence/Hugging-Face?id=23410

Important note about your webcam

Your USB webcam on the PC is fine. ARC can use that with the Camera Device skill. The webcam is not the issue here.

Make sure you have:

1. Camera Device skill added
2. Webcam selected in that skill
3. Camera skill started
4. Hugging Face skill configured to use that camera feed

If the camera is not running, Hugging Face would usually complain about no active camera, not model support.

Best next step

Try this:

1. Open the Hugging Face skill settings
2. Switch to Object Detection
3. Use the default detection model
4. Press Save
5. Run Query Now

If that works, then the issue is specifically the Qwen vision-chat model/provider combo.

If you want Vision Chat

If your goal is the "robot sees and talks" behavior, you may need to:

• choose a different vision model that your enabled provider supports, or
• use a Hugging Face Inference Endpoint / compatible provider that supports that model

Quick question

Are you trying to do:

• face detection / tracking, or
• vision chat where the model describes the scene and answers questions?

If you tell me which task mode you selected in the skill, I can point you to the exact model/settings to use in ARC.