ZoomGain-4B-GGUF

GGUF-format weights for ZoomGain-4B, converted with llama.cpp.

• Language model: FP16, Q8_0, Q4_K_M.
• Vision encoder and projector (mmproj): FP16, Q8_0, Q4_K_M.

For image input, provide both the language model and the corresponding mmproj file.

CLI Inference (llama-mtmd-cli)

llama-mtmd-cli \
  -m ZoomGain-4B-Q8_0.gguf \
  --mmproj mmproj-ZoomGain-4B-Q8_0.gguf \
  --image /path/to/image.jpg \
  -p "Describe this image." \
  --temp 0 -n 2048

Web Chat (llama-server)

llama-server \
  -m ZoomGain-4B-Q8_0.gguf \
  --mmproj mmproj-ZoomGain-4B-Q8_0.gguf \
  --host 127.0.0.1 --port 8080

Replace Q8_0 with F16 or Q4_K_M to use another precision. Use a llama.cpp build with Qwen3.5 multimodal support. Quantization can affect accuracy.