Hacker News new | ask | show | jobs
Langport-A self-hosted LLM serving platform that provides OpenAI-compatible APIs (github.com)
3 points by JasonWang1027 1124 days ago
1 comments

I'm excited to share with you a new open-source project called langport, which aims to provide a lightning-fast large language model serving platform.

Inspired by lmsys/fastchat, we've built a distributed LLM serving system. Our focus on performance means we use batch inference to get higher throughput on the serving platform. Langport offers a range of core features, including streaming API interface support, batch inference for higher throughput, and OpenAI-Compatible RESTful APIs.