PrismML Deploys 1-Bit AI Models to Qualcomm Smart Glasses

The 2-billion-parameter Bonsai LLM represents a significant compression feat, shrinking standard model architectures by a factor of four while maintaining benchmark performance. By running locally on Qualcomm-powered hardware, the model allows users to query their surroundings through vision and language, effectively turning smart glasses into contextual assistants.

Founded by Caltech researchers and advised by UC Berkeley’s Ion Stoica, PrismML advocates for open-weight AI that prioritizes on-device efficiency over the resource-heavy requirements of proprietary labs. While the technology is now verified for Qualcomm’s architecture, no consumer hardware featuring the integrated model has reached the market yet.