Google DeepMind has announced updates to its Gemini 3.5 Transcribe model, enhancing its speech-to-text capabilities. The updated model demonstrates improved accuracy in recognizing complex data like phone numbers and order IDs, even in noisy conditions. It also features enhanced language support, transcribing speech in over 85 languages, and can now remove filler words and auto-format text. Developers can integrate these improvements into their applications via the Gemini app on macOS and Gboard on Android, with further access through Google AI Studio. AI
IMPACT Enhances speech-to-text capabilities for developers and users, improving accuracy and language support in transcription services.
RANK_REASON Google DeepMind announced updates to its Gemini 3.5 Transcribe model. [lever_c_demoted from frontier_release: ic=2 ai=1.0]
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →