PrismML deploys its compact language models on Qualcomm-enabled smart glasses
At Qualcomm's Snapdragon Summit, PrismML demonstrated its 1-bit Bonsai LLM running locally on smart glasses powered by the Snapdragon AR1 Gen 1 platform. The 2-billion-parameter model, compressed 4x while retaining near-original benchmark performance, handles real-time vision and language queries entirely on-device. Founded by Caltech researchers and advised by UC Berkeley's Ion Stoica, PrismML aims to advance open-weight, privacy-focused AI that keeps data local, though no shipping products have been announced yet.
PrismML, an AI lab founded by Caltech researchers and advised by UC Berkeley's Ion Stoica, has adapted its compact language models to run directly on smart glasses powered by Qualcomm's Snapdragon chips. The move matters because it signals a push toward AI that works entirely on-device—no cloud connection, no constant data uploads—right inside the hardware wearable makers already use.
What Happened at Snapdragon Summit
On Wednesday, at Qualcomm's Snapdragon Summit, the chipmaker demonstrated PrismML's 1-bit Bonsai LLM running locally on AI smart glasses built on the Snapdragon AR1 Gen 1 Platform. The demo showed a model small enough to fit the tight power and thermal constraints of eyewear, yet capable of handling real questions from the wearer.
How Bonsai Gets So Small
PrismML's core technique, as TechCrunch has previously reported, is compressing larger models by a substantial factor—in this case, 4x—while keeping nearly all of their performance on standard benchmarks. The smart glasses version is a 2-billion-parameter model tuned for both vision and language. That combination means a wearer can ask, in real time, about whatever they are looking at, and get an answer processed on the glasses themselves.
Why On-Device AI Is the Point
The company's broader ambition is open-weight AI that runs on devices and makes better use of the computing power consumers already own. PrismML frames this as an alternative to relying on the privacy promises of proprietary AI labs—and on their seemingly endless appetite for more compute. A model living on the device keeps visual and voice data local by default.
Porting Bonsai to Qualcomm's chip is a concrete step toward that vision. What to watch next: no smart glasses actually shipping with PrismML's model have been announced yet, so the next milestone will be a hardware partner putting it into a real product.
Meta description: PrismML's 1-bit Bonsai LLM now runs locally on Snapdragon AR1 Gen 1 smart glasses, bringing private, on-device AI vision and language to wearables.
Tags: PrismML, Qualcomm, Snapdragon AR1 Gen 1, Bonsai LLM, on-device AI
Featured image: Abstract technology style—miniature glowing tree structure inside a translucent lens against a dark circuit backdrop, no people, no logos, no text.
Comments
No comments yet. Be the first to comment.