Gemini Docs

Audio transcription

• New guide: Audio transcription — speech-to-text with Gemini 3.5 Transcribe (gemini-3.5-transcribe), including automatic language detection (85+ locales), speaker diarization, word-level timestamps, custom vocabulary, and smart transcription.
• New guide: Audio transcription (generate content) — same model via generate_content / Files API upload examples (Python, JavaScript, REST).
• New Live API guide: Live transcription with Gemini Live API — real-time streaming STT with gemini-3.5-transcribe-live over WebSockets / the Gen AI SDK.
• New model page: Gemini 3.5 Transcribe — unary (gemini-3.5-transcribe) and Live (gemini-3.5-transcribe-live) variants, limits, and feature matrix.

Models & Live API

• Models: Gemini 3.5 Transcribe listed as a new stable speech-to-text model (gemini-3.5-transcribe, gemini-3.5-transcribe-live).
• Live API: use cases and key features updated for live transcription/captioning and continuous speech-to-text with language detection and custom vocabulary.

Pricing

• Pricing: paid-tier rates documented for Gemini 3.5 Transcribe and Gemini 3.5 Transcribe Live (audio input / text output; free tier unchanged at no charge).

Changelog & deprecations

• Changelog: August 26, 2026 entry — Gemini 3.5 Transcribe and Transcribe Live marked generally available (GA).
• Deprecations: gemini-3.5-transcribe, gemini-3.5-transcribe-live, and gemini-3.5-live-translate-preview added to the model tables (no shutdown date announced).