User Guide

This section provides comprehensive documentation for using vLLM Ascend in production environments.

Support Matrix

• Supported Models — Models supported by vLLM Ascend
• Supported Features — Features supported on Ascend NPU
• Feature Matrix — Feature compatibility across hardware versions

Configuration

• Environment Variables — Configure vLLM Ascend via environment variables
• Additional Configuration — Additional configuration options

Feature Guide

Explore detailed guides for vLLM Ascend features including graph mode, CPU binding, quantization, sleep mode, structured output, LoRA, expert parallelism load balancing, and more.

Deployment Guide

• Using Volcano Kthena — Deploy with Volcano Kthena
• Using MindIE Motor — Deploy with MindIE Motor

Release Notes

• Release Notes — Latest release information