Scribe v2

Audio modelCard summarized

Scribe v2 is built for batch transcription, subtitling, and captioning at scale

● Re-checked 12 Oct 2026 · 45 facts · 0 corrections · 5 unconfirmedSource: Introducing Scribe v2 launch post and Speech to Text docs (no model card published)

Use it via API

$0.22 / hour

Code example Compare

Where to call it

ElevenLabs APIscribe_v2

Supports

No feature list in the maker's docs.

Full feature support

Good for

Batch transcription of long and complex recordings
Subtitles and captions with word-level timestamps
Meeting documentation with speaker diarization

Overview

ElevenLabs · 01Scribe

Scribe v2

Scribe v2 is built for batch transcription, subtitling, and captioning at scale

Typical price

$0.22 / hour

Scribe v2 (batch)

Max length: 10 h input (3 GB)
Languages: 90+ languages
Modes: speech-to-text, diarization (up to 32 speakers)
Released: 9 Jan 2026

Scribe v2 is ElevenLabs' batch speech-to-text model for transcription, subtitling and captioning at scale. ElevenLabs says it improves on Scribe v1's stability and accuracy on long-form audio, pauses, tone changes and long silences, and adds keyterm prompting, entity detection with timestamps and automatic multi-language transcription. launch post

scribe_v2

Typical price: $0.22 / hour
Max length: 10 h input (3 GB)
Languages: 90+ languages
Weights: Closed
Released: 9 Jan 2026

speech-to-text
diarization (up to 32 speakers)
word timestamps
audio event tagging
keyterm prompting
entity detection
multichannel (up to 5)