Microsoft is expanding the reach of local AI capabilities in Windows 11 by enabling developers to run language models on PCs without a dedicated NPU, instead leveraging Nvidia GPUs. This experimental update to the Windows App SDK, spotted by Windows Latest and reported by TechRadar, allows non-Copilot+ devices with a GeForce RTX 30 series or newer GPU (with at least 6GB of video RAM) to run local AI tasks such as text rewriting or summarization.
Technical Details of the Windows App SDK Update
- Microsoft's Windows App SDK now includes experimental Language Model APIs that can run on non-Copilot+ PCs.
- The requirement: a supported Nvidia GPU (RTX 30 series or newer with 6+ GB vRAM).
- This enables local execution of Microsoft's Phi Silica small language model without an NPU.
- Previously, such local AI features were restricted to Copilot+ PCs with a fast neural processing unit.
| Feature |
Sources:
