PrismML releases tiny language models for Qualcomm smart glasses
PrismML unveiled a version of its small language models designed to run on Qualcomm-powered smart glasses at the Snapdragon Summit on Wednesday.
Model size and performance goals
The startup demonstrated a one-bit Bonsai LLM that fits onto devices using the Snapdragon AR1 Gen 1 Platform. This specific model contains two billion parameters and is optimized for both vision and language tasks. PrismML states that this approach shrinks larger models by four times while keeping performance high on standard tests.
Local processing and privacy focus
The company aims to create open-weight AI that utilizes the computing power already present in user devices. This strategy offers an alternative to relying on proprietary labs that require massive amounts of compute resources. The release marks a step toward running AI locally without depending on external cloud promises.
Current availability status
Although the software is ready for Qualcomm chips, no physical smart glasses running this model have been announced yet. The demonstration took place at the Snapdragon Summit event held on Wednesday. PrismML continues to advise its work alongside researchers from UC Berkeley and Caltech.
Reported by one outlet
Only one outlet has published this. Nothing here has been checked against a second report, so read it as that outlet's account and follow the link below for the original.
Reported by
1 independent outlet. Headline as published. Links open the original report.