NVIDIA's Project G-Assist was announced last year, and we went hands-on with a preview version at Computex 2024. The AI-powered assistant for GeForce RTX gaming rigs is designed to offer real-time performance analysis and optimization and Wiki-style assistance for gameplay, optimizing builds, or finding out more about what's on screen and what you're playing. It's fascinating tech, and it's going to be released in Beta form very soon.

With the GeForce RTX 50 Series getting its formal reveal at CES 2025 and the flagship GeForce RTX 5090 and RTX 5080 set to launch at the end of January, NVIDIA has confirmed that the new G-Assist System Assistant feature is launching into beta sometime in February 2025.
It's set to become available as an add-on within the NVIDIA App, and its features will be accessible via a chat or voice interface within the app's overlay. G-Assist System Assistant runs on a relatively small Llama-based Instruct model with 3 billion parameters so that it can run natively with minimal latency on a range of GeForce RTX hardware, not just the new GeForce RTX 50 Series.
- Read more: NVIDIA Project G-Assist - your in-game RTX AI Assistant for guides, tips, and overclocking
- Read more: Seeing Project G-Assist in action made us want NVIDIA's AI Assistant for all our PC gaming
So then, what can it do? Well, G-Assist System Assistant can provide real-time feedback on a game's performance, optimize settings, provide information on potential bottlenecks, and even interact with hardware to overclock your GPU, adjust fan curves, and even make changes to lighting on compatible Logitech G, Corsair, MSI, and Nanoleaf devices.
Here's an example of the AI-powered G-Assist System Assistant in action in Remedy's Alan Wake 2.
After being asked to analyze performance, G-Assist gives a detailed summary of the average FPS, GPU utilization, the display's refresh rate, and the current display mode, noting that the performance is GPU-limited. This makes sense for a visually intensive game like Alan Wake 2 with Path Tracing enabled.
Here's another example, also in Alan Wake 2. The G-Assist System Assistant provides a detailed CPU and GPU utilization chart similar to what you might find in a thorough analysis or review that requires spreadsheets, tables, and custom plotting. Whereas that might take someone considerable time and effort to put together, the G-Assist System Assistant can whip up performance charts in seconds.
It's impressive, opening the door to an effortless and natural way to benchmark performance and tweak system and in-game settings. As G-Assist is launching into beta, we probably won't see one of the technology's most fascinating aspects and potential uses. And that's the ability for it to provide feedback on the game you're playing, with recommendations on what items to equip in an RPG while also answering questions on what you're seeing on-screen.
These features are expected to arrive later. G-Assist is built on NVIDIA ACE, the same technology being used to create AI NPCs and digital humans for games. NVIDIA is opening up access to G-Assist through GitHub, and companies like MSI and Logitech are already using the technology to integrate it into their software and apps for gamers and content creators.