Vision / Image support

[grigio]

it worked on OpenWebUI+ollama but not in OpenWebUI+llama-swap+llamacpp

500: Failed to parse messages: Unsupported content part type: "image_url";

ggml-org/llama.cpp#12348

[mostlygeek]

When llama-server supports vision/multi-modal models llama-swap will be support it as well. For now you can use an another inference server with vision support. I use vllm and it works through llama-swap.