PrismML's 1-bit Bonsai LLM runs on Qualcomm AR1 smart glasses — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- PrismML demonstrated its 1-bit Bonsai LLM running locally on smart glasses built on Qualcomm's Snapdragon AR1 Gen 1 Platform at the chipmaker's Snapdragon Summit on Wednesday.
- The startup's core claim is that its models shrink larger models by 4x while retaining almost all performance on standard benchmarks, as previously reported by TechCrunch.
- Qualcomm's Snapdragon AR1 Gen 1 chip runs a 2-billion-parameter version of PrismML's model tuned for vision and language, letting wearers ask about what they see in real time.
- PrismML, founded by Caltech researchers and advised by UC Berkeley's Ion Stoica, frames its work as open-weight on-device AI — an alternative to proprietary labs' privacy promises and insatiable compute demands.
- No smart glasses running PrismML have been announced yet; the Qualcomm chip demonstration is a step toward that goal but no commercial product has materialized.
Why it matters: The demo validates that a 2-billion-parameter 1-bit vision-language model can run on AR glasses silicon in real time, undercutting the assumption that smart glasses must lean on cloud AI. For Qualcomm, the AR1 Gen 1 platform gains a local-LLM showcase ahead of any commercial PrismML-powered glasses — and for privacy-sensitive wearers, on-device inference removes a key data-residency concern.
Ask SkimNews
