Gemini 3.5 Transcribe
Speech-to-text - gemini-3.5-transcribe
- Price (provider's unit)
$2.00 per 1M audio input tokens ($0.003/min); $12.00 per 1M text output tokens ($0.002/min)
“| Input price | Free of charge | $2.00 or $0.003/min* (audio) | | Output price (including thinking tokens) | Free of charge | $12.00 or $0.002/min* (text) |”
ai.google.dev - Converted $/hour (our arithmetic)
- Not convertedPriced per token
- Languages
85+ languages
“| Language Auto-detection | Supported (85+ languages) | Supported (85+ languages) | Includes mid-session code-mixing. |”
ai.google.dev - Streaming
No
“For real-time, low-latency streaming transcription over WebSockets, visit the Live transcription guide using gemini-3.5-transcribe-live.”
ai.google.dev - Max audio
1 hour per request (30 minutes with diarization or word timestamps)
“Up to 1 hour per request (up to 30 minutes when speaker diarization or word-level timestamps are enabled)”
ai.google.dev - Status
stable (no shutdown date)
“| gemini-3.5-transcribe | August 2026 | No shutdown date announced |”
ai.google.dev
Features
speaker diarization (up to 8 speakers)
“| Speaker Diarization | Not Supported | Supported (up to 8 speakers) |”
ai.google.dev word-level timestamps
“| Word-level Timestamps | Not Supported | Supported |”
ai.google.dev custom vocabulary (up to 1,000 terms)
“| Custom Vocabulary Biasing | Supported (up to 1,000 terms) | Supported (up to 1,000 terms) |”
ai.google.dev
Google states an effective blended rate of ~$0.005 per min (25 audio tokens per second, 175 text tokens per minute).