Your version.
Your docs.
Start building with vLLM Agentic API. Choose the documentation that matches the version you run.
Rust command-line reference
Latest / development · The agentic CLI: run harnesses, serve, and validate.
Python command-line reference
Latest / development · The agentic-api Python launcher: commands, options, and defaults.
Start here
An introduction to the project and its documentation.
API reference
Explore the API surface for your selected version.
Codex Desktop
Use the desktop UI with local models: tested Linux setup and tool limitations.
Developing
Set up a development environment and work on the project.
Contributing
Find ways to participate and contribute to the project.
SGLang upstream
Connect SGLang and review the tested configuration.
NVIDIA Dynamo upstream
Connect a Dynamo deployment with a vLLM worker.
agentic-llm-d
Explore the v1alpha response hydration and persistence design.