8 GB accelerator memory
Q4 / mobile quant · llama.cpp / MLX
The small Gemma 4 tier is designed for laptop and edge deployment; long multimodal context still adds memory.
Text model
A compact multimodal model designed for everyday hardware.
Capabilities
The most useful reasons to choose this model, without making you read through its repository first.
Everyday assistants
Document understanding
Low-cost multimodal apps
For providers
The provider requirements live here, separate from the information you need to choose and use the model.
8 GB accelerator memory
Q4 / mobile quant · llama.cpp / MLX
The small Gemma 4 tier is designed for laptop and edge deployment; long multimodal context still adds memory.
16 GB system RAM
Q4 GGUF · llama.cpp
CPU serving is plausible at short context, but throughput must be measured locally.
Source and license
OpenMayhem links back to the source repository so you can inspect the model card, files, license, limitations, and creator guidance directly.