PrismML Adapts 2-Billion-Parameter Bonsai Model for Qualcomm Smart Glasses
The Caltech-founded startup showcased its compressed 1-bit vision-language model running locally on Snapdragon AR1 hardware.
PrismML, an artificial intelligence startup founded by Caltech researchers and advised by UC Berkeley professor Ion Stoica, has created a version of its lightweight models engineered to run directly on smart glasses powered by Qualcomm silicon.
At Qualcomm’s Snapdragon Summit on Wednesday, the chipmaker demonstrated PrismML’s 1-bit Bonsai model operating locally on smart glasses built on the Snapdragon AR1 Gen 1 Platform, according to reporting by TechCrunch AI (https://techcrunch.com/2026/09/24/prismml-brings-its-tiny-llms-to-qualcomm-powered-smart-glasses/).
The smart glasses release is a 2-billion-parameter vision and language model designed to let wearers query visual inputs in real time without routing processing through remote servers. PrismML’s core technology compresses larger models by a factor of four while preserving standard benchmark performance.
The startup positions its open-weight, on-device models as an alternative to proprietary cloud-based AI providers, arguing that localized execution eliminates both continuous data transmission risks and reliance on centralized server compute.
While the summit demonstration verifies execution on Qualcomm's AR platform, no commercial smart glasses featuring PrismML models have been announced to date.
Sources
Written by
The Company Wire
Inside the companies building what’s next. Reporting on startups, technology, funding and the people shaping them.
