flashinfer is an open-source LLM inference tool written in Cuda, released under the Apache-2.0 license. The repository has collected 6.7k stars and 1.6k forks; the latest push was in October 2026. Docs and downloads are linked from flashinfer.ai.
“FlashInfer: Kernel Library for LLM Serving”
— flashinfer-ai, project description on GitHub
Details
| Category | Local LLMs & Inference |
|---|---|
| GitHub stars | 6,690 |
| Forks | 1,609 |
| Language | Cuda |
| License | Apache-2.0 |
| Last updated | |
| Owner | flashinfer-ai |
| Website | flashinfer.ai |
| Repository | flashinfer-ai/flashinfer |
Tags
attentioncudadistributed-inferencegpujitlarge-large-modelsllm-inferencemoe
Repository facts from the GitHub public API, refreshed October 18, 2026. The quoted tagline is written by the project owner; the summary above is generated from public facts.