v0.23.1
📦 ollamaView on GitHub →
✨ 1 features🐛 1 fixes🔧 2 symbols
Summary
This release introduces Gemma 4 MTP speculative decoding support for Macs, significantly boosting performance for the Gemma 4 31B model, alongside general threading fixes and a Go version bump.
Migration Steps
- To use the new MTP speculative decoding for Gemma 4 31B, run: `ollama run gemma4:31b-coding-mtp-bf16`
✨ New Features
- Gemma 4 MTP (Multi-token Processing) speculative decoding is now supported on Macs, potentially yielding over a 2x speed increase for the Gemma 4 31B model on coding tasks.
🐛 Bug Fixes
- Threading fixes applied to MLX and MLX-C.