v0.32.4

wxt-dev/wxtv0.32.4Jul 25, 2026by github-actions[bot]

AI Summary

This release focuses on GPU support, model handling improvements, and fixing decoding issues across various architectures.

Key Highlights

Added support for Apple GPUs via MLX engine for models like Laguna
Improved quantization for draft-model output heads
Fixed Qwen3 MoE decoding for differently-quantized experts

Full Release Notes

## What's Changed
- Support Laguna on Apple GPUs via the MLX engine
- Quantize draft-model output heads at the requested type when creating speculative-decoding drafts.
- Fixed Qwen3 MoE decoding for differently-quantized experts, plus faster packed gate/up projection (~4–9% on M5 Max).

**Full Changelog**: https://github.com/ollama/ollama/compare/v0.32.3...v0.32.4