PrismML runs 2B‑parameter vision‑language model on Snapdragon smart glasses
Demo shows compressed, open‑weight models handling real‑time visual queries fully on-device, but no products are announced yet.
PrismML has adapted its ultracompact 1‑bit Bonsai language model to run fully on-device on Qualcomm Snapdragon–based smart glasses, demonstrated at the Snapdragon Summit. The showcased model is a 2‑billion‑parameter vision‑language system that can answer questions about what the wearer is seeing in real time. By compressing larger models by about 4x while keeping nearly the same benchmark performance, PrismML enables richer AI experiences on low‑power AR hardware. The company is pursuing open‑weight, on-device AI as an alternative to cloud‑hosted proprietary models that require heavy compute and raise privacy concerns. Although Qualcomm demonstrated the technology on its Snapdragon AR1 Gen 1 Platform, no commercial smart glasses products using PrismML have been revealed yet.
Why it matters
This demo suggests smart glasses could move from simple heads-up displays to assistants that understand what the wearer sees and answer questions on the spot, without sending data to the cloud. PrismML’s compression technique promises near‑full model performance on low‑power Snapdragon hardware, and its open‑weight stance offers an alternative to relying on proprietary cloud AI providers. Commercial devices are still unannounced, so for now this is a signal of where AR hardware may be headed rather than a feature users can buy.
Signal or noise?
Does this story matter, or is it hype? Decide before you see what everyone else thinks.
Sources
- TechCrunch