Local AI with Gemma 4 inference snap
Gemma 4 E4B is Google's efficient multimodal instruction-tuned model with vision capabilities.
Use this snap to quickly install an optimized environment for local inference with Gemma 4.
The snap includes the following hardware-optimized inference engines:
- cpu: CPU-optimized for workstations
- nvidia-gpu: CUDA-optimized for workstation GPUs
- amd-gpu: ROCm-optimized for workstation GPUs
- intel-cpu: Optimized for Intel CPUs using OpenVINO Model Server
- intel-gpu: Optimized for Intel GPUs using OpenVINO Model Server
- intel-onemkl: Intel CPU-optimized for workstations using oneMKL (experimental)
The most suitable engine is automatically selected based on the available hardware.
Install:
sudo snap install gemma4
Run:
gemma4
Some accelerators require extra drivers to be usable with this snap:
https://documentation.ubuntu.com/inference-snaps/how-to/setup/drivers/
License:
The Gemma 4 model is provided by Google under the Apache 2.0 license:
https://ai.google.dev/gemma/apache_2
The licenses of all the bundled software can be found after installing the snap at
/snap/gemma4/current/usr/share/doc.