PrismML adapts its compact AI models for Qualcomm-powered smart glasses
PrismML has built a version of its compact language model that runs on Qualcomm Snapdragon chips for smart glasses. The model was showcased at Qualcomm's Snapdragon Summit, though no glasses using it have been announced yet.
PrismML, a startup founded by Caltech researchers with UC Berkeley's Ion Stoica serving as an advisor, has developed a version of its small language model designed for smart glasses powered by Qualcomm's Snapdragon chips.
On Wednesday, at Qualcomm's Snapdragon Summit, the chipmaker demonstrated PrismML's 1-bit model called Bonsai, which can run locally on AI smart glasses built on the Snapdragon AR1 Gen 1 Platform.
Smaller size, retained performance
PrismML has previously become known for its ability to substantially shrink large language models — in this case by a factor of four — while keeping nearly all of their original performance on standard benchmarks. The glasses version is a 2-billion-parameter model tuned for both vision and language, allowing wearers to ask questions in real time about what they're looking at.
PrismML's broader ambition is to advance open-weight AI that runs directly on devices, making fuller use of the computing power those devices already contain. The company frames this approach as an alternative to relying on the privacy promises of proprietary AI labs and their ever-growing demand for computing resources.
Releasing a model for Qualcomm's chip marks a step toward that goal, though no specific smart glasses product using PrismML's technology has been announced so far.
