NVIDIA has unveiled “Chat with RTX,” a trailblazing application that empowers users to create their very own local AI chatbot, reminiscent of ChatGPT, by leveraging the power of NVIDIA’s graphics processing units (GPUs).
Chat with RTX, currently in its demo phase, is designed to simplify the creation of AI chatbots, enabling users to harness the capabilities of their NVIDIA-equipped PCs. Its minimalist interface belies the complex technology under the hood, making it accessible to users of all technical backgrounds. Whether you’re sifting through documents for specific data or summarizing YouTube videos, the application promises to streamline these tasks with unprecedented ease.
Users can choose between two large language models (LLMs) to power their chatbot: Mistral, from the French startup of the same name, or LLaMA 2, developed by Meta. This choice allows users to tailor the chatbot’s capabilities to their specific needs.
Chat with RTX requires a PC with an NVIDIA GPU from the RTX 30 or RTX 40 series, equipped with at least 8 GB of VRAM. Other system requirements include Windows 11, a minimum of 16 GB of memory RAM, and the latest version of the graphics card drivers. The application’s performance and capabilities highlight the need for a powerful PC setup, with the software taking up about 40 GB of storage space and utilizing a significant amount of RAM during operation.
What sets Chat with RTX apart is its customizable nature. Unlike other AI models that offer broad functionalities, Chat with RTX requires personal training. Users can feed it text documents, notes, videos, and more, tailoring the AI’s knowledge base to their specific needs. Once trained, Chat with RTX can perform tasks like summarizing extensive PDFs, analyzing data to draw conclusions, or even answering questions based on a YouTube playlist.
NVIDIA’s Chat with RTX diverges from the cloud-based models that giants like Google, Microsoft, and Meta have developed, offering a unique, locally-run alternative that promises to revolutionize how we interact with AI.
