Nvidia is adding resource controls that let RTX Spark laptop owners decide how GPU capacity is split between an active application and an AI agent running in the background. The setting appears under System and Advanced in the Nvidia App and requires Windows hardware-accelerated GPU scheduling.
The update addresses a practical conflict on machines used for both ordinary graphics work and local AI: a persistent agent and a foreground app can compete for the same compute. Users can favor the application for responsiveness or assign more resources to the agent so it finishes sooner. Nvidia has not described the feature as automatic scheduling; it is a user-controlled allocation.
RTX Spark systems combine a Blackwell RTX GPU and Grace CPU with up to 128GB of unified memory and are available from Asus, Dell, HP, Lenovo, Microsoft and MSI. Nvidia also announced support for Liquid AI’s LFM3 model family and its new Nemotron 3.5 Transcribe speech model. That model can transcribe 40 languages and combine transcription, speaker labels, translation and questions about recorded audio. Model availability and performance depend on the specific laptop configuration and software format.