Newly released open-weight models like OpenAI’s gpt-oss and Alibaba’s Qwen 3 can run directly on PCs, offering high-quality outputs for local agentic AI applications. NVIDIA RTX PCs accelerate these experiences, providing fast and snappy AI to users interested in exploring generative AI applications locally.

NVIDIA has optimized top LLM applications for RTX PCs, extracting maximum performance of Tensor Cores in RTX GPUs. Ollama, an open-source tool, simplifies running and interacting with LLMs on PCs, supporting features like drag-and-drop PDFs, conversational chat, and multimodal workflows with text and images.

Enthusiasts can also use LM Studio, an app powered by llama.cpp, to run models locally, chat with them in real time, and integrate them into custom projects. NVIDIA has optimized performance on RTX GPUs for LM Studio, supporting the latest NVIDIA Nemotron Nano v2 9B model and offering CUDA kernel optimizations for faster inference.

Local LLMs offer greater privacy and performance, enabling context-aware AI conversations and personalized study assistants. AnythingLLM allows students to create interactive flashcards, quizzes, and guides from study materials, providing adaptive, interactive study companions on RTX PCs for faster responses and customization.

Project G-Assist, an AI assistant for gaming PCs, has received an update introducing commands to adjust laptop settings like app profiles optimized for laptops, BatteryBoost control, and WhisperMode control. Users can extend functionality through the G-Assist Plug-In Builder and discover new plugins via the G-Assist Plug-In Hub for expanded capabilities.

Read more at Nvidia: How to Get Started With Large Language Models