Skip to content

An editorial team and a community building with AI

vibecoding.incJoin Discord ↗
← All articles

News · Google · 2 min read

Event date: · Published:

Google released Gemini 3.5 Transcribe speech-to-text models for recorded and live audio

On August 26, 2026, Google released Gemini 3.5 Transcribe for pre-recorded audio and Gemini 3.5 Transcribe Live for real-time streaming, both with speaker attribution, word-level timestamps and custom vocabulary.

Archive entry: this covers an event from August 26, 2026, and we published it on November 8, 2026. The details come from sources dated at the time and may have changed since.

On August 26, 2026, Google released Gemini 3.5 Transcribe for pre-recorded audio and Gemini 3.5 Transcribe Live for real-time streaming transcription. The Google blog post on Gemini 3.5 Transcribe said both models offered speaker attribution, word-level timestamps and custom vocabulary.

What changed

  • There were two models: gemini-3.5-transcribe (non-streaming) and gemini-3.5-transcribe-live (bidirectional WebSocket streaming through the Live API).
  • Features included speaker diarization, word-level timestamps and a custom vocabulary of up to 1,000 terms.
  • Word error rate on Artificial Analysis was 4.0% for streaming and 2.6% for non-streaming. On FLEURS it was 5.50% for streaming and 5.04% for non-streaming.
  • Google said time to final transcription was 70% better than Chirp 3.
  • Pre-recorded audio got speaker attribution with timestamps for up to three speakers, and support for more than three speakers was experimental. Files up to 1 hour long were processed through the Interactions API.

Availability and language coverage

The blog said the models transcribed over 85 languages, while the Gemini API changelog entry for August 26, 2026 said 80+. The changelog listed the models as generally available, while the blog described developer availability as public preview in the Gemini API and Google Antigravity. Google said the models powered Rambler on Gboard and dictation in the Gemini app for macOS, and were coming to Chrome.

What it means for people who build with AI

Developers got Google's own streaming and batch transcription with diarization in the Gemini API, competing with dedicated speech-to-text services. On September 15, 2026, Google also published a post on building real-time voice applications with Gemini 3.8 Live and 3.5 Transcribe.

Sources

Written with AI assistance, strictly from the sources listed above. Before it went live, automated checks verified its dates, sources and every language edition.

Google released Gemini 3.5 Transcribe speech-to-text models for recorded and live audio | vibecoding.inc