Google DeepMind introduced SL2T, a model that translates sign language into text, on August 12, 20261. It powers sign-to-text dictation in Gboard and Live Transcribe on Pixel 11, starting with American Sign Language (ASL) to English1. The company positions this as bringing sign language AI out of the lab and into consumer products for the first time1. Training data covers more than 100,000 hours across over 50 sign languages, roughly a quarter of it in ASL — but what ships today is ASL to English only1.
The architecture is the notable part. SL2T treats sign language as a sequence of pose landmark locations rather than a raw camera feed1. An on-device model (MediaPipe Holistic) tracks the location of points on the signer, and only those geometric coordinates are sent to the server for translation, allowing the original video to be discarded immediately1. Alongside the growing number of small models that run on the device itself, it is an implementation that splits the work between on-device preprocessing and server-side inference. There is no additional cost, and more devices are said to be coming1.
Sources
- Putting sign language AI into users’ hands - Google DeepMind official blog (August 12, 2026)