Lizard AI Runtime, also referred to as Lizard LLM, is a hardware-aware local AI runtime application designed for Windows 10 and Windows 11. Its primary function is to enable users to run large language models (LLMs) entirely offline on their own personal computers. The software automatically inspects the system's processor, memory, and graphics hardware to recommend a compatible GGUF model. Once a model is selected and set up, all prompts, responses, and model weights remain on the device, and inference operates without any internet connection. The application includes two native inference engines: Lizard Native, which uses Direct3D 12 for inference on compatible resident graphics cards, and Caterpillar, which is designed for CPUs and shared or integrated graphics. Lizard automatically selects the appropriate engine for the machine and can fall back to bundled llama.cpp or an optional Ollama installation when needed. The application provides a local chat interface for direct interaction with the AI model. It also includes an OpenAI-compatible API, allowing other software applications to connect to and utilize the local model. Activation requires a one-time payment of five dollars, with no subscription fees. The tool is intended for users who require private, offline AI assistance, such as for sensitive data processing, work without internet access, or personal experimentation with local models. The typical workflow involves installing the Windows application, allowing Lizard to inspect the hardware, selecting a recommended model from the provided list, and then running private AI chat or connecting third-party software through the local API. Technical documentation, installation guides, API references, benchmark methodology, and answers regarding privacy, pricing, GPU compatibility, and offline use are available on the product website.