NVIDIA cuts VRAM requirement to run G-Assist, almost all RTX GPUs now have access

NVIDIA has announced it is expanding the number of devices that can access its intuitive, locally run AI, Project G-Assist, with the company now lowering the barrier of entry through a new AI model.

NVIDIA is kicking off Gamescom 2025 by announcing a new AI model for Project G-Assist, which uses 40% less VRAM. As a result, RTX GPUs with 6GB or more VRAM, including laptops, can now access the virtual assistant. In addition to lowering the barrier of entry for Project G-Assist, NVIDIA has laptop commands that users can take advantage of.

With G-Assist now available in a broader range of RTX PCs, including laptops, NVIDIA has informed us that, despite requiring 40% less VRAM, G-Assist will maintain its accuracy in responses. Furthermore, G-Assist has received its biggest update yet, as users are now able to access the G-Assist Plug-In Hub, as NVIDIA has partnered with mod.io.

The partnership enables G-Assist users to easily access G-Assist plug-ins, which are created by the community. Moreover, G-Assist can discover, download, and install plug-ins directly from the chat window through the use of natural language. NVIDIA explained to us that G-Assist will automatically detect the right plug-in a user requests, download it, and inform the user how to operate it.

Notably, NVIDIA recommends 12GB of VRAM to run the current G-Assist model, which showcases how much of a substantial upgrade this is.