Particle.news

Google Unveils Gemini 3.5 Live and Transcribe

The updates aim to make voice interactions more natural by handling interruptions, processing live visuals, switching languages, improving transcription, rolling out first to developers with phased consumer updates to select apps.

Overview

  • Google announced three Gemini 3.5 variants — 3.5 Live, 3.5 Live Experimental, and 3.5 Transcribe — and made them available to developers in public preview through the Gemini API and AI Studio.
  • Consumer rollout has begun in English for the macOS Gemini app and the Rambler dictation feature on Android in selected countries and languages, with Chrome support promised soon.
  • Gemini 3.5 Live is designed to handle mid-sentence interruptions, accept live visual input, switch between languages on the fly, and trigger background tools to make voice chats more natural and responsive.
  • Gemini 3.5 Live Experimental narrates its step-by-step reasoning for complex tasks, while Gemini 3.5 Transcribe formats text automatically, removes filler words, accepts custom vocabularies, timestamps words, and attributes speech for up to three speakers.
  • Google says 3.5 Transcribe is a major advance over its prior Chirp 3 model, a change that could speed note-taking, improve meeting transcripts and help multilingual users, while the promised 3.5 Pro variant has not yet been released.