Google Upgrades Fitbit Air Workout Detection With LLM Model
Google says the new LLM-driven motion detection model delivers 200% more accurate activity classification and cuts missed workouts in half.
Google announced a software upgrade for the Fitbit Air that brings a new motion detection model to the device's automatic workout detection, according to Android Authority. The update targets how the screenless tracker recognises and classifies exercise without user input.
The new model is designed to improve detection for activities that were previously harder to identify, including swimming and weight training. Google says it now uses LLM reasoning to combine location and routine data with movement data when classifying an activity.
According to Google, the change results in 200% more accurate classification and half as many missed workouts. The company says better workout detection was the most requested feature from Fitbit Air owners.
Screenless trackers
The Fitbit Air is part of a category of screenless activity trackers, alongside devices such as WHOOP, that have been gaining popularity with users who value simplicity and long battery life, as noted in October 2025 coverage of the segment.
Quick answers
What does the Fitbit Air update change?
It introduces a new LLM-driven motion detection model for automatic workout detection, improving classification of activities such as swimming and weight training.
How much more accurate is the new detection?
Google claims the improvement delivers 200% more accurate classification and half as many missed workouts.
Why did Google focus on workout detection?
Google says better workout detection was the most requested feature from Fitbit Air owners.