At CES, NVIDIA announced the launch of foundation models running locally on NVIDIA RTX AI PCs, enhancing digital humans, content creation, productivity, and development. These models, available as NVIDIA NIM microservices, are accelerated by the newly introduced GeForce RTX 50 Series GPUs. These GPUs, built on the NVIDIA Blackwell architecture, boast up to 3, 352 trillion operations per second of AI performance and 32GB of VRAM, supporting FP4 compute for improved AI inference performance. This support enables generative AI models to operate locally with a smaller memory footprint compared to previous hardware. The GeForce platform has historically been crucial for AI developers; over 30% of recent AI research papers referenced GeForce RTX usage. With the rise of generative AI and RTX AI PCs, new low-code and no-code tools make AI development accessible to more users. NIM microservices and NVIDIA AI Blueprints provide straightforward and preconfigured workflows for digital applications. Leading PC manufacturers are launching NIM-ready RTX AI PCs featuring these innovations. NVIDIA also introduced a pipeline of NIM microservices for RTX AI PCs, developed by top model developers like Black Forest Labs and Meta. These microservices support various applications across language models, image generation, PDF extraction, and computer vision.

Robin Rombach, CEO of Black Forest Labs, emphasized the performance improvement and broader access provided by these advancements. Additionally, NVIDIA revealed the Llama Nemotron family of models for agentic AI tasks. The Llama Nemotron Nano, available as a NIM microservice, excels at tasks like instruction following and chat. NIM microservices work seamlessly across NVIDIA GPUs, making AI deployment straightforward. Enthusiasts and developers can easily download and run these microservices on Windows 11 PCs using the Windows Subsystem for Linux (WSL). NVIDIA is optimizing these microservices for Windows PCs, ensuring accessibility for developers building AI features into their applications. The compatibility of these microservices with major AI development frameworks makes it easier for developers to leverage the latest technologies. The Project R2X preview demonstrates how developers can create AI-enabled avatars using NVIDIA’s tools. NVIDIA's generative AI capabilities, including new AI Blueprints, allow users to execute complex workflows such as converting PDFs into podcasts or guiding image creation with 3D assets. Starting in February, NVIDIA NIM microservices and AI Blueprints will be available for selected NVIDIA GPUs, with broad availability promised across various PC manufacturers. These developments highlight NVIDIA's leadership in the rapidly evolving field of AI-enabled computing, offering a flexible and powerful platform for both developers and enthusiasts.