Gemini 3.5 Transcribe

Gemini 3.5 Transcribe (google/gemini-3.5-transcribe) — Google’s speech recognition model for transcribing recordings: lectures, interviews, meetings. Works via POST /v1/audio/transcriptions, billed per minute of audio (from actual duration).

Parameters

Field	Description
model	google/gemini-3.5-transcribe
file	Audio file (mp3, wav, webm, m4a etc.); up to 25 MB
language	ISO-639-1 code (ru, en…) — improves accuracy; auto-detected when omitted

Pricing

$0.0058 per minute of audio — charged by actual duration. An hour of recording ≈ $0.35.

Example

curl https://api.mixen.ai/v1/audio/transcriptions \
  -H "Authorization: Bearer $MIXEN_API_KEY" \
  -F model=google/gemini-3.5-transcribe \
  -F language=ru \
  -F file=@meeting.mp3

Strengths and limitations

• No diarization (who spoke) and no word-level timestamps — text only; segment on your side.
• Robust to noise and conversational speech; the language hint noticeably improves accuracy on short recordings.
• Files over 25 MB — split before the request.

All models — in the catalog. The full transcription workflow — guide.