Multimodality? #2
Description
Does this have any features for merging/routing models based on modalities (text, image, embedding, transcription, etc.)? It's not clear to me from the docs I've seen so far, but this would be a great feature, as most local models aren't "omni" like 4o, so emulating those needs a frontend that can merge.