GM
Gemma 4 12B, 26B, and 31B
FeaturedWorkstation-class Gemma 4 models for stronger local reasoning, coding, multimodal input, and long context.
Workstation Models#Gemma 4#workstation model#reasoning#multimodal
Gemma 4 Workstation Models
Gemma 4 workstation models are good choices when the user has a high-memory Mac, a 24 GB GPU, or a larger local workstation.
Recommended Sizes
| Model | Package size | Context | Good for |
|---|---|---|---|
| Gemma 4 12B | 7.6 GB | 256K | Fast workstation assistant |
| Gemma 4 26B | 18 GB | 256K | MoE local reasoning and vision |
| Gemma 4 31B | 20 GB | 256K | Strong dense local model |
Install Commands
ollama run gemma4:12b
ollama run gemma4:26b
ollama run gemma4:31b
Hardware Notes
The 12B model is the practical starting point. Use 26B or 31B when you have 32 GB plus unified memory, 24 GB plus VRAM, or a server-class setup.