Find tools › Open-source projects › Keep data private

ds4

Open-source project · by antirez

Run high-end DeepSeek and GLM AI models locally on your Mac or multi-GPU PC.

Some setupPopular

What it costs

Free and open source; you may pay for the AI model it uses.

Use it to

  • Run DeepSeek V4 locally on Apple Silicon or CUDA GPUs
  • Host a local OpenAI-compatible server for coding tools
  • Run an integrated local coding agent from the command line

How to start

  1. Clone the repository and enter the directory with cd ds4.

  2. Compile for your hardware using make, make cuda-generic, or the appropriate target.

  3. Download a supported model using ./download_model.sh ds4f-q2.

  4. Start interacting with the model by running ./ds4 or ./ds4-server.

Good to know. The project is rapidly changing and in beta quality; it is intentionally narrow and requires specific GGUF models produced for this project rather than general GGUF files.

Before you install

  • Found automaticallyNot reviewed by a person yet. Check it before installing.
  • License: MITYou are allowed to use it under these terms.
  • Last updated 2026-11-08

More detail

A specialized native C inference engine and server optimized for running DeepSeek V4 and GLM models on consumer hardware, Macs, and CUDA systems.

General GGUF runners can be slow or lack custom optimizations for specific giant MoE models. Reach for ds4 to run models like DeepSeek V4 Flash or PRO on Apple Silicon Macs, DGX Spark, or multi-GPU CUDA setups.

Similar tools

Descriptions are written in plain English from each project's own documentation and refreshed by our agent; this one was last checked 2026-10-04. Spotted a mistake? · Open in Find tools · © 2026 Which AI Map