It’s OK
The appeal of this app is that you have one place to use both on-device and remote AI.
However, the on-device support is hobbled by the inability to associate a downloaded GGUF wile with an “mmproj” GGUF file to enable multimodal (i.e., image) input. So your manually-downloaded models cannot interpret attached pictures, whether from the Photos app or your camera.
However, the on-device support is hobbled by the inability to associate a downloaded GGUF wile with an “mmproj” GGUF file to enable multimodal (i.e., image) input. So your manually-downloaded models cannot interpret attached pictures, whether from the Photos app or your camera.
Show more
Response from developer
Thank you for the feedback! You’ve raised a good point. I hadn’t considered importing a separate mmproj file for manually added models. I’m currently working on an update, so I’ll look into whether I can include this feature.
