HLVM routes through a selected model. Local default models use the managed runtime; cloud models use configured provider credentials.
hlvm model list
The list groups models by provider and marks local/cloud availability.
hlvm model set ollama/gemma4:12b
hlvm model set claude-code/claude-sonnet-4-6
The setting is shared by the REPL, hlvm ask, and runtime chat APIs.
hlvm model pull ollama/gemma4:12b
Local pulls go through HLVM's managed runtime. Do not start another Ollama
instance on 11434 to make a model appear.
hlvm model key list
hlvm model key set openai
hlvm model key unset openai
Provider keys are stored in the platform credential store where available.
The current source tier map selects:
| Host memory | Model |
|---|---|
>= 16 GiB | gemma4:12b |
< 16 GiB | gemma4:e2b |
The tier map lives in src/hlvm/runtime/embedded/model-tiers.json.