BitNet

microsoft/BitNet

Inference framework for 1-bit LLMs

PersonalEnterpriseLocal inferenceLibrary / SDKPermissive
github.com
Preview of BitNet
Preview of BitNet

About

Microsoft's official runtime for fast 1-bit models on CPUs.

From the repository: “Official inference framework for 1-bit LLMs”

llama.cppLLM inference in C/C++ on CPUs and GPUsCLILibrary / SDK86
vLLMHigh-throughput LLM serving engineDocker / self-hostLibrary / SDK86
UnslothFast local fine-tuning and running of LLMsWebLibrary / SDK81
SGLangFast serving framework for LLMsDocker / self-hostLibrary / SDK81