Skip to content
vllm.cpp vllm.cpp

Documentation

  • Quickstart
  • Usage
  • Building
  • Features
  • Benchmarks
  • Environment variables
  • Binary releases
  • Speculative decoding
  • KV offload
  • Weight offload
  • SGLang compatibility
  • ROCm backend
  • Kolibri1 TT captured decode gap analysis
github.com/mudler/vllm.cpp
Built from docs/ at HEAD.

docs/README.md

Reference pages

Use these pages when you need exact interface behavior, values, or fields.

  • C API
  • Engine lifecycle
  • Model loading
  • Server

The existing public references remain the primary owners for these topics:

  • Benchmarks
  • Build options
  • Environment variables
  • Features
  • Release artifacts
  • Status

For runnable commands, start with the generic workflows in USAGE.md .