microsoft/Tutel
Microsoft's optimized Mixture-of-Experts implementation library for LLM inference and training with advanced quantization support.
Not currently ranked — collecting fresh signals.
star history
Tutel is a performance-optimized library for Mixture-of-Experts architectures in large language models. It provides parallel solutions including No-penalty Parallellism for models with dynamic behaviors during training and inference. The library supports FP8, NVFP4, MXFP4, and BlockwiseFP8 quantization formats across CUDA and ROCm GPUs for serving MoE-based models including DeepSeek, Kimi, Qwen3, GLM-5, and GptOSS.
Frequently asked
- What is microsoft/Tutel?
- Microsoft's optimized Mixture-of-Experts implementation library for LLM inference and training with advanced quantization support.
- Is Tutel open source?
- Yes — microsoft/Tutel is open source, released under the MIT license.
- What language is Tutel written in?
- microsoft/Tutel is primarily written in C.
- How popular is Tutel?
- microsoft/Tutel has 999 stars on GitHub.
- Where can I find Tutel?
- microsoft/Tutel is on GitHub at https://github.com/microsoft/Tutel.