GM
Gemma 4 Thinking Models
Gemma 4 models with configurable thinking behavior for local reasoning, coding, agentic tasks, and multimodal work.
Reasoning Models#Gemma 4#thinking mode#reasoning#local model
Gemma 4 Thinking Models
Gemma 4 is useful when the user wants reasoning behavior, multimodal input, and local workstation options in one model family.
Recommended Sizes
- Gemma 4 E4B for compact local testing.
- Gemma 4 12B for a practical workstation default.
- Gemma 4 31B when local quality matters more than memory cost.
Install Commands
ollama run gemma4:e4b
ollama run gemma4:12b
ollama run gemma4:31b
Hardware Notes
Reasoning mode, long context, and multimodal input increase runtime cost. Leave memory headroom instead of choosing the largest model that barely loads.