ZhiLight is an open-source LLM inference tool written in C++, released under the Apache-2.0 license. It has 907 GitHub stars and 109 forks, and was last updated in March 2026. Related topics: cuda, deepseek-r1 and gpt.
“A highly optimized LLM inference acceleration engine for Llama and its variants.”
— zhihu, project description on GitHub
Details
| Category | Local LLMs & Inference |
|---|---|
| GitHub stars | 907 |
| Forks | 109 |
| Language | C++ |
| License | Apache-2.0 |
| Last updated | |
| Owner | zhihu |
| Repository | zhihu/ZhiLight |
Tags
cudadeepseek-r1gptinference-enginellamallmllm-inferencellm-serving
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.