GLaDOS

GLaDOS is a local AI voice assistant modeled on the sarcastic character from Valve's Portal games. It's for people who want a conversational companion on their own hardware, with camera awareness and connections to home automation. The Python project is open source under the MIT license and runs on Linux and Windows. macOS support is experimental.

It can start conversations without a wake word, responding to camera events, sounds or time. You can interrupt its speech, and your input takes priority over background activity. FastVLM handles scene understanding and change detection, while Parakeet recognizes speech. Kokoro and a custom GLaDOS speech model provide spoken responses; a terminal interface also accepts text.

Its personality extends beyond a prompt. Separate agents contribute memory, mood and observations without blocking the main conversation. Long-term memory retains facts, preferences and conversation summaries across sessions. PAD models reactive mood, while HEXACO traits keep the character consistent. You can change its personality and voice.

The language model can run locally through Ollama, including Llama 3.2 or Mistral, or use an OpenAI-compatible cloud API such as MiniMax or OpenRouter. Speech and vision models run locally; choosing a cloud LLM sends language-model requests to that provider. CPU-only operation works but is slower, and GPU options include NVIDIA, AMD and Intel.

MCP tools connect it to Home Assistant and expose system information. The project also provides an OpenAI-compatible speech API backed by Kokoro, which can run in Docker.