First announced and shown off at GTC earlier in March, the update adds several new capabilities:

• ChatRTX adds to its growing list of supported LLMs, including Gemma, Google’s latest LLM, and ChatGLM3, an open, bilingual (English and Chinese) LLM, providing users with additional flexibility.
• New photo support enables ChatRTX users to easily search and interact locally with their own photo data without the need for complex metadata labeling, thanks to OpenAI’s Contrastive Language-Image Pre-training (CLIP). 
• ChatRTX users can now speak with their own data, with added support for Whisper, an AI automatic speech recognition system that now enables ChatRTX to understand verbal speech. 

ChatRTX enables users to unleash the power of AI on their RTX-powered PCs today, leveraging the accelerated performance of TensorRT-LLM software and NVIDIA RTX. And it's all done locally on your PC, so your data is secure. Moreover, it's available on GitHub as a free reference project, enabling developers to explore and expand AI applications using RAG technology for even more diverse use cases.

Chat with RTX, now free to download, is a tech demo that lets users personalize a chatbot with their own content, accelerated by a local NVIDIA GeForce RTX 30 and 40-Series GPU with at least 8GB of video random access memory, or VRAM.