LM Studio 0.4.13 Released: Performance Boost for Qwen and Gemma on Mac

LM Studio released version 0.4.13 on May 13. The update focuses on performance improvements for specific AI models on Mac systems.

Parallel prediction speeds up vision models

The update includes mlx-engine version 1.8.1 for Mac. Performance has been improved for Qwen 3.5, Qwen 3.6, and Gemma 4 vision models through parallel prediction. A bug that caused newlines to collapse when pasting into chat input has been fixed.

The update is available globally as of May 13. LM Studio recommends the update for all users due to security enhancements.

The new version targets users running large language models locally. The parallel prediction feature aims to speed up vision-capable models. LM Studio has not confirmed further details about future updates.