HQLHLVM DocsHQL DocsGitHub
Download
HLVM for macOSmacOS 26+ · .dmg

First Run

HLVM starts with auto. It never downloads, installs, or starts Ollama, Python, or a model during installation or launch.

hlvm ask "summarize the main points of REST in five bullets"

What Happens

  1. HLVM starts its lightweight local runtime host.
  2. The core discovers compatible installed agent runtimes, their authentication state, configured provider credentials, and any existing Ollama service.
  3. auto selects an eligible concrete route for the request.
  4. The selected route streams the answer through the same HLVM conversation.

If at least one route is ready, no setup screen or command is required.

If Authentication Is Needed

hlvm doctor
hlvm model list
hlvm model login <backend>

Use the exact backend shown by hlvm model list. A native app may present the same focused sign-in action. HLVM does not infer a paid subscription without an installed authenticated vendor runtime or supported login flow.

If No Route Is Available

Install and authenticate one supported agent runtime or configure a provider key. Existing Ollama installations are optional and discovered automatically; HLVM does not install or own Ollama.

To opt into a local model after installing and starting Ollama yourself:

hlvm model pull ollama/qwen3:8b
hlvm model set ollama/qwen3:8b

The pull is always explicit. Local Ollama is never an injected auto fallback.

Diagnose

hlvm doctor
hlvm model list
hlvm ask --verbose "hello"

Next