LocalAI

LocalAI is a self-hosted local AI engine that provides APIs compatible with OpenAI. After installation, you can select models in the Web interface for scenarios such as chat, text to speech, speech recognition, and vector embedding. For scripts, tools or automation that are already connected to the OpenAI API, you only need to point the base URL to LocalAI on this lazy cat micro service to switch to the local model. This lazy cat micro-service version uses the official LocalAI CPU image and does not require a graphics card. Novices can start testing with small models from 0.5B to 4B, and then try larger models or image or voice models after confirming that the process is running. Models, inference back-end, configuration and build files are saved in the persistence directory and remain after restarts and upgrades. LocalAI does not pre-install models when it starts. Users can install models in the built-in model market; if models cannot be downloaded directly on the current network, they can also manually download the `.gguf` file and then import it through Import Model / Advanced YAML.