# ROCm CLI

A command-line tool for setting up and running local AI models on AMD GPUs, shipped as a single prebuilt binary with a full-screen TUI dashboard for GPU telemetry, model serving and chat.

ROCm CLI ships as a single prebuilt binary for Linux and Windows, needing no existing Python, Rust or ROCm install, and bundles inference-engine adapters for Lemonade and vLLM. It can inspect the local environment, discover available engines and models, find a running service, and chat with a locally served model.

Its live dashboard, with GPU telemetry, model serving status and chat, requires Linux or WSL2; on Windows only the CLI and Lemonade-based serving are available, without the live dashboard or vLLM support. The minimum supported base is Ubuntu 24.04 or an equivalent glibc 2.38 with `GLIBCXX_3.4.32`, which the Lemonade engine is linked against.

The project is explicitly marked as a tech preview, with APIs, commands and behavior that may change without notice; an installer script downloads a bundle, verifies its SHA-256 checksum, and installs `rocm` and `rocmd` into `~/.local/bin`.

Website: https://rocm.ai