Google Translate adds model switcher, letting users prioritize speed over accuracy

Google Translate is set to unveil a model switcher that allows users to choose between faster translation and better accuracy, according to 9to5Google. Currently, the model picker has ‘Fast’ and ‘Advanced’ selections, with the latter being the default option. Users may find the switch button below the Google Translate logo.

Google calls these additions ‘text-translation models,’ with the Advanced version offering ‘high accuracy for complex translations.’ Users now have the option to switch to the ‘Fast’ alternative in the Google Translate app, which is described as being ‘best for quick translation.’ While 9to5Google spotted this model picker on the iOS version of the app, it has not yet appeared for all users, including the Android Central team on either the iOS or Android applications.

The selection of text translation models applies only to typed translation requests, as indicated by the name. At the bottom of the model switcher, there is a disclaimer that states: “Advanced is supported for text translation only in select languages.” It is unclear which languages are currently supported or if access to the model picker is limited to paid Google AI Pro or Google AI Ultra subscribers.

Since the model switcher is designed exclusively for text requests, it will not impact live translation—a feature that is becoming increasingly popular across various translation tools and brands. Users of Google Pixel Buds Pro 2 and Pixel Buds 2a can benefit from live translation; however, they do not have the option to manually select the translation model that powers this feature.

When available, Google Translate app users can tap the pill-shaped model indicator to switch between Fast and Advanced. By default, users will still get the most accurate translation model, but they now have the option to prioritize speed for certain situations.

Google Translate app users can tap the pill-shaped model indicator to switch between Fast and Advanced when available. By default, users will still get the most precise translation model, but they now have the option to prioritize speed for certain situations.