AI lab PrismML, founded by Caltech researchers and advised by UC Berkeley’s Ion Stoica, has developed a version of its tiny language models designed to run on smart glasses powered by Qualcomm’s Snapdragon chips.
Qualcomm showcased PrismML’s 1-bit Bonsai LLM at its Snapdragon Summit last week, demonstrating that the model can run locally on AI glasses built on the Snapdragon AR1 Gen 1 platform. The smart glasses version is a 2-billion-parameter model tuned for vision and language, allowing wearers to ask in real time about what they are looking at.
PrismML is known for shrinking larger models substantially, in this case by four times, while retaining nearly all of their performance on standard benchmarks. Its broader aim is open-weight AI that runs directly on devices and makes better use of the computing power they already contain. The startup positions this approach as an alternative to relying on the privacy promises of proprietary AI labs and their ever-growing demand for compute.
The Qualcomm release marks a step toward that vision, although no smart glasses running PrismML’s model have yet been announced.