Gemma 3 on Ollama

Min VRAM: 4 GB

Pull

ollama pull gemma3

Run

ollama run gemma3

API

After running, Ollama serves an OpenAI-compatible API at http://localhost:11434/v1.

Model card