User Guide

Install, deploy, and run UCM. Start with the installer, choose a model family, then follow the serving-engine and deployment guidance for your environment.

Sections

• Installation — select a published artifact and get the matching command
• Getting Started — vLLM, vLLM Ascend, and SGLang integration
• Support Matrix — supported models, platforms, and feature coverage
• Deploy — container and Kubernetes deployment
• Deployment Frameworks — deploy with pyMotor or Kubernetes
• Model Tour — model-family catalogs and engine-specific launch guidance
• Capabilities — core capabilities including Prefix Cache
• Observability — metrics and monitoring
• Diagnostics — diagnostic tools including Trace Mode