llama.cpp thinks it's not a vision model?
I downloaded Qwen 3.5 from another repo and it was recognized as a vision model, but the one from here wasn't. I simply loaded the model in llama.cpp via the command:llama-server -hf mradermacher/Qwen3.5-4B-PaperWitch-heresy-i1-GGUF:Q6_K --jinja -c 0 --host 127.0.0.1 --port 8033
Edit, another model of yours: mradermacher/Ministral-3-8B-Reasoning-2512-absolute-heresy-i1-GGUF:Q4_K_M also wasn't recognized as a vision model.
So I'm guessing I'm doing something wrong then? I'm new llama.cpp, I apologize if it's obvious.
For vision capability you need in addition to the LLM stack to provide the mmproj file containing the vision stack available in the static repository:
Had trouble getting it recognized with llama.cpp, but it worked fine with Jan, which uses llama.cpp in the background anyway. Good enough for me. Thank you.