Back to list
ai-toolsLLMinferenceMetalMacDeepSeeklocal
ds4 - DeepSeek V4 Flash Local Inference Engine for Metal
A native Metal inference engine for DeepSeek V4 Flash by antirez (Redis creator). Runs 284B parameter model on 128GB MacBooks with 2-bit quantization, SSD-persistent KV cache, and 1M token context.
3 views1605 stars5/8/2026
A native Metal inference engine for DeepSeek V4 Flash by antirez (Redis creator). Runs 284B parameter model on 128GB MacBooks with 2-bit quantization, SSD-persistent KV cache, and 1M token context.