How to use from
llama.cppInstall from WinGet (Windows)
winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf 1bit-MONSTER/gemma-3-1b-it-GGUF:Q4_K_M# Run inference directly in the terminal:
llama cli -hf 1bit-MONSTER/gemma-3-1b-it-GGUF:Q4_K_MUse pre-built binary
# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf 1bit-MONSTER/gemma-3-1b-it-GGUF:Q4_K_M# Run inference directly in the terminal:
./llama-cli -hf 1bit-MONSTER/gemma-3-1b-it-GGUF:Q4_K_MBuild from source code
git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf 1bit-MONSTER/gemma-3-1b-it-GGUF:Q4_K_M# Run inference directly in the terminal:
./build/bin/llama-cli -hf 1bit-MONSTER/gemma-3-1b-it-GGUF:Q4_K_MUse Docker
docker model run hf.co/1bit-MONSTER/gemma-3-1b-it-GGUF:Q4_K_MQuick Links
gemma-3-1b-it — GGUF (Q4_K_M)
llama.cpp project's own Q4_K_M GGUF, re-hosted with measured performance for the 1bit engine on Strix Halo.
Licensed under Google's Gemma Terms of Use.
Contents
gemma-3-1b-it-Q4_K_M.gguf
Measured performance (Strix Halo, Vulkan)
pp512: 5791 tok/s · tg128: 83.3 tok/s
Running it
1bit serve -m gemma-3-1b-it-Q4_K_M.gguf --device vulkan
Attribution
- Base model: google/gemma-3-1b-it.
- Quantization: ggml-org/gemma-3-1b-it-GGUF.
- Downloads last month
- 76
Hardware compatibility
Log In to add your hardware
4-bit
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Install (macOS, Linux)
# Start a local OpenAI-compatible server with a web UI: llama serve -hf 1bit-MONSTER/gemma-3-1b-it-GGUF:Q4_K_M# Run inference directly in the terminal: llama cli -hf 1bit-MONSTER/gemma-3-1b-it-GGUF:Q4_K_M