Fact · Speech recognition
Google describes Gemini 3.5 Transcribe as its speech-to-text model, with automatic language detection, speaker diarization, word-level timestamps and custom vocabulary biasing.
Checked against its source on October 17, 2026
In the source's own words
“Our speech-to-text model with automatic language detection, speaker diarization, word-level timestamps, and custom vocabulary biasing.”
Gemini Developer API pricing, Google, Google AI for Developers, 2026, Model list, entry for Gemini 3.5 Transcribe.
Open the original- What it measures
- This is the model entry on Google's Gemini Developer API price list, read on 2026-10-17. It sits beside the Gemini 3.5 Transcribe Live entry, which streams audio.
- What it does not show
- This is Google's own description of the model's features. The page gives no accuracy figures, so it does not show how well the model recognizes speech.