返回首页
AI on Medium··行业媒体

How I Serve Qwen3.8–27B on an RTX 5090: vLLM, Headless Ubuntu, and Real-World Benchmarks

中文摘要

本指南介绍如何在 Ubuntu Server 24.04 上使用 vLLM 在单张 RTX 5090 (32GB) 部署 Qwen3.8-27B 及相关基准测试。

English Summary

This guide details deploying Qwen3.8-27B on a single RTX 5090 (32GB) using vLLM on headless Ubuntu Server 24.04, including real-world benchmarks.

原文节选

In this guide, I show how I deployed Qwen3.8–27B on a single RTX 5090 32 GB using vLLM on headless Ubuntu Server 24.04, exposed it through… Continue reading on Medium »