PrismML launched compact LLMs on smart glasses with Qualcomm chips

PrismML's Tiny-LLMs land on smart glasses with Qualcomm chips

👓

PrismML has brought its compact language models to smart glasses with Qualcomm chips. The models are open-weight and run on-device: inference happens directly on the device, without offloading computation to the cloud.

Announcement from September 24, 2026, details and the accompanying image — techcrunch.com

Wearable gadgets are the tightest spot for LLMs in terms of hardware: a cloud runtime won't fit here, so all the weight falls on the chip.

Open-weight + on-device — these are models that compute locally and don't hit a server on every request.

PrismML's stated goal is to squeeze more out of the compute hardware already in the device