Project G-Assist
AI agent · Computer-use / browser agent by Nvidia
AI assistant to optimize and control your GeForce RTX PC.
Project G-Assist is a local AI assistant for GeForce RTX PCs that optimizes system performance and settings via voice or text commands. It runs on the GPU, offering privacy and responsiveness, and is extensible with community plugins.
Category: AI Agents
- Platforms
- Windows 10, Windows 11
- Languages
- English
- Operating System
- Windows 10, Windows 11
- Disk Space Required (System Assistant)
- 7 GB
- Disk Space Required (Voice Commands)
- 3 GB
- Driver Requirement
- GeForce 572.83 driver, or later
- AI assistant runs locally on GeForce RTX GPUs for responsive, offline use.
- Optimizes system performance, game settings, and peripherals via voice or text commands.
- Extensible with community-developed plugins for custom functionality.
- Uses a Llama-based Instruct model with 8 billion parameters for efficient local AI processing.
Overview
Project G-Assist is an AI assistant designed for GeForce RTX PCs that helps users tune, control, and optimize their systems. It runs locally on the user's GeForce RTX GPU, ensuring responsiveness, offline functionality, and privacy, unlike cloud-hosted AI models. Powered by a Llama-based Instruct model with 8 billion parameters, G-Assist interprets voice or text commands to perform tasks such as maximizing system performance, optimizing power efficiency, adjusting game settings, providing real-time diagnostics, and controlling select peripherals. It can be integrated with agentic frameworks and extended through community-developed plugins, offering a customizable and powerful tool for PC users.
Core Capabilities
- Local Processing: Runs locally on GeForce RTX GPUs, offering responsiveness and offline capabilities.
- Underlying Model: Uses a Llama-based Instruct model with 8 billion parameters for language understanding.
- GPU Resource Allocation: Allocates a portion of GPU horsepower for AI inference, with potential for brief dips in performance during use.
- Plugin Extensibility: Supports community-driven expansion through plugins, allowing developers to add new functionality.
- Agentic Framework Integration: Can be integrated with agentic frameworks like Langflow for use in larger AI pipelines and multi-agent systems.
- System Diagnostics and Optimization: Provides real-time diagnostics and recommendations to alleviate system bottlenecks and improve power efficiency.
- System Control and Tuning: Can adjust game settings, overclock the GPU, and control select peripherals and software applications.
- Reasoning Mode: Offers new Reasoning Mode for enhanced accuracy and intelligence, enabling multiple actions from a single prompt.
- Flash Mode provides faster responses compared to previous releases.
Extensibility and Development
- Plug-In Builder: Users can create and customize G-Assist's functionality using a ChatGPT-based Plug-In Builder.
- Custom Plug-in Development: Developers can define functions in JSON formats and write logic in Python to integrate new tools or features.
- Developer Resources: NVIDIA provides a GitHub repository with sample plugins, instructions, and documentation for developing custom functionalities.
- Plugin SDKs: Supports Python, C++, and Node.js plugins for various development needs.
User Interaction
- NVIDIA Overlay Integration: Users can interact with G-Assist directly from the NVIDIA overlay without tabbing out of applications.
- Voice and Text Commands: Supports voice or text commands for system control and optimization.
Use cases
- Optimize PC performance and power efficiency
- Adjust game and system settings
- Receive real-time system diagnostics and performance metrics
- Control select peripherals and software applications
- Customize system functionality with plugins
- Integrate with larger AI pipelines and multi-agent systems
Specifications
- Platforms
- Windows 10, Windows 11
- Languages
- English
- Operating System
- Windows 10, Windows 11
- GPU Requirement
- GeForce RTX 30, 40, and 50 Series Desktop GPUs with 12GB VRAM or Higher
- CPU Requirement
- Intel Pentium G Series, Core i3, i5, i7, or higher; AMD FX, Ryzen 3, 5, 7, 9, Threadripper or higher
- Disk Space Required (System Assistant)
- 7 GB
- Disk Space Required (Voice Commands)
- 3 GB
- Driver Requirement
- GeForce 572.83 driver, or later
- Underlying Model
- Llama-based Instruct model with 8 billion parameters
- Processing Environment
- Local processing on GeForce RTX GPU
- Extensibility
- Supports community-driven plugins via GitHub repository
- Integration
- Can integrate with agentic frameworks like Langflow
How it works
- Initiate Command: Users can interact with G-Assist through voice or text commands, or via the NVIDIA overlay.
- AI Inference: G-Assist processes the command using a local Llama-based model, allocating GPU resources as needed.
- Execute Action: G-Assist executes actions such as tuning system settings, adjusting game options, or providing diagnostics.
- Extend Functionality (Optional): Users can expand G-Assist's capabilities by developing and integrating custom plugins.
FAQ
What is Project G-Assist and how does it work?
Project G-Assist is an AI assistant that runs locally on your GeForce RTX GPU, meaning it is responsive, free to use, and can run offline. It uses a Llama-based Instruct model with 8 billion parameters.
What can Project G-Assist do?
Project G-Assist can help tune and optimize your system, maximize performance, optimize power efficiency, adjust game settings, provide real-time diagnostics, chart performance metrics, and customize peripherals.
What technology powers Project G-Assist?
Project G-Assist uses a Llama-based Instruct model with 8 billion parameters. When prompted, your GeForce RTX GPU briefly allocates a portion of its horsepower to AI inference.
Can Project G-Assist be extended with new features?
Yes, G-Assist was built for community-driven expansion. NVIDIA has published a GitHub repository with samples and instructions for creating plugins that add new functionality.
What are the system requirements for Project G-Assist?
Project G-Assist requires a GeForce RTX 30, 40, or 50 Series Desktop GPU with 12GB VRAM or higher, running Windows 10 or 11, with at least 7GB of disk space for the System Assistant and 3GB for Voice Commands. A GeForce 572.83 driver or later is also required.