DeepGEMM
DeepGEMM is a clean and efficient GPU library of BLAS kernels for developers working on GPU linear algebra. DeepGEMM: clean and efficient BLAS kernel library on GPU
Pricing
Open source (free)
Update history
Last updated 2026-11-19;Listed 2026-11-16
Author
deepseek-ai
More from this author
Trust info
Listed on 2026-11-16; Last updated 2026-11-19; Data synced on 2026-11-21
Trial feedback
Like it? Vote to help it rank higher.
From the README
DeepGEMM is a unified, high-performance tensor core kernel library that brings together the key computation primitives of modern large language models — GEMMs (FP8, FP4, BF16), fused MoE with overlapped communication (Mega MoE), MQA scoring for the lightning indexer, HyperConnection (HC), and more —…
- Stars
- ★ 9.2k
- Category rank
- #464
- Language
- Cuda
- License
- MIT
- Last updated
- 2026-11-19
- Category
- Developer tools
- Section
- General





