DFlash MLX Server macOS ARM64 (1936b7a)

DFlash MLX Server — macOS ARM64

Built from main branch at commit 1936b7a.

What's included

• mlx-server — standalone OpenAI-compatible inference server
• Standard MLX inference (mlx-lm)
• DFlash speculative decoding acceleration

Usage

tar -xzf dflash-mlx-server-macos-arm64.tar.gz

# Standard MLX inference
./mlx-server --model /path/to/model --port 8080

# DFlash accelerated inference
./mlx-server --model /path/to/model --draft-model /path/to/draft --port 8080

For Atomic Chat integration

This binary is automatically downloaded during the Atomic Chat build process.