12-vision-multimodal

Documentation

Overview

Recipe 12: Vision / Multimodal Inference

Analyze images using a vision-capable GGUF model. The image is passed alongside a text prompt using the inference.Message API, the same format used by the OpenAI-compatible /v1/chat/completions endpoint.

Requirements:

• A vision-capable GGUF model (e.g. LLaVA, Gemma 3 with vision encoder)

Usage:

go run ./docs/cookbook/12-vision-multimodal/ --model path/to/vision-model.gguf --image photo.jpg
go run ./docs/cookbook/12-vision-multimodal/ --model path/to/vision-model.gguf --image photo.jpg --prompt "Count the objects"