Ollama v0.31.1 Release: Faster Gemma 4 on Apple Silicon

Core summary

One sentence to understand this update

Ollama v0.31.1 significantly boosts Gemma 4 performance on Apple Silicon, achieving nearly 90% faster token generation in a coding-agent benchmark by utilizing multi-token prediction (MTP).

Impact & opportunity

What this could mean

Apple Silicon users working with Ollama will experience substantial speed improvements for Gemma 4, making local development and deployment of agents more efficient.