UnslothAI releases the 4-bit MLX version DGEMMA 4-31B optimized for Apple Silicon

2026-04-11 00:40:19

Abstract generation in progress

ME News Report, April 11 (UTC+8), UnslothAI recently released the native 4-bit MLX optimized version of the DGEMMA 4-31B model. This version is specifically designed for Apple Silicon and is claimed to achieve extremely fast inference speeds on all M-series Macs while maintaining efficient memory usage (about 20GB RAM). Additionally, the model features powerful multimodal and visual capabilities, supporting a full 256K context length and native function calling features. (Source: InFoQ)

View Original

This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.

1 Likes