using multimodal models in llama-server

[deven367]

Hello,
Thank you for this amazing project! I am currently using the b6969 build with gemma-3-4b on my m2 pro macbook, the new web-ui is unable to accept images/pdfs as input

I'm running the server as follows

./build/bin/llama-server -m ~/Downloads/gemma-3-4b-it-Q5_K_M.gguf

Does the cli need any additional commands for multimodal models to work? I'm aware of the llama-mtd-cli, however that's not the command I wish to use. Can someone please help me with this?

Thanks!

[abc-nix]

You are missing the mmproj gguf.

[deven367]

Ah yes, after posting the question, I had figured it out! Thank you so much! :)