Phi-3 on Ollama

Min VRAM: 4 GB

Pull

ollama pull phi3

Run

ollama run phi3

API

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

Model card