Home-LLM
A local LLM integration for Home Assistant with voice, chat and AI automations. Runs on Raspberry Pi without a GPU and supports Ollama and llama.cpp.
Home-LLM connects Home Assistant to language models running on your own hardware, so you can control smart devices through voice or chat. It's for Home Assistant users who want natural language control without relying on a cloud service or subscription. The project pairs a custom integration with small models trained specifically for smart home commands.
Its Home models include Home-Llama-3.2-3B and Home-FunctionGemma-270m. They turn requests into Home Assistant actions and can run on modest hardware, including Raspberry Pi, without a GPU. You can also use instruction-tuned models such as Qwen3, Llama 3 and Mistral, with examples supplied by the integration to guide smart home control.
Models can run directly within Home Assistant through llama.cpp or on a separate machine through Ollama and llama.cpp server. OpenAI-compatible connections support LM Studio, LocalAI and vLLM; text-generation-webui is another supported backend. Local processing keeps your data on your hardware. Optional cloud connections through OpenAI Responses or Anthropic Messages APIs send data to the provider, including the state of exposed devices and potentially your location.
The integration handles lights, locks, climate controls, media players and vacuums, along with timers, todo lists and scripts. It connects to Home Assistant's speech-to-text and text-to-speech pipelines and supports streaming voice responses. AI tasks can extract data and generate structured responses or dynamic content for automations. Built-in language support covers English, German, French, Spanish and Polish.
