[FEATURE] Support for more multimodal models on Ollama

[kk-spartans]

Is your feature request related to a problem? Please describe.

Would be nice if we could any multimodal model from ollama, especially models with more parameters. Llava 7b is pretty dumb.

Describe the solution you'd like

Support for other models, for example: "operate -m llava:32b"

[ghaz1]

I agree especially that now we have gemma3:4b which is way smarter and better performance

[joshbickett]

@ghaz1 good point. I'm going to revisit and review this soon