Gemini 3.5 Transcribe Accelerates AI Speech-to-Text Across Google Apps
Google unveils Gemini 3.5 Transcribe, a faster (about 70% quicker) and more accurate AI voice-to-text model that cleans up disfluencies like ums and uhs, supports custom vocabulary, and works in 85 languages for up to three speakers. Initially live in Gboard’s Rambler on Pixel 11, it will expand to more Gemini devices and the Chrome browser soon, with macOS voice input benefiting today. Developers can access the Gemini API, and Google is integrating the tech with partners like Antigravity and AI Studio.