[Feat]: Support for multimodal models (e.g., Phi-4-multimodal) #230
Description
Description
With the advent of multimodal modals like Phi-4-multimodal (5.6b parameters), it might be interesting to enable interacting with these models in the app.
Use Case
Would be great to upload an image and talk about it. As Phi-4-multimodal also support speech, that could also be included in the feature request. However, I think images might have a higher priority, for instance for visually impaired pocketpal users who could start relying on offline on-device image recognition and descriptions.