ElevenLabs launches Scribe v3 and Scribe v3 Live speech-to-text models

ElevenLabs released Scribe v3 and a new realtime variant, Scribe v3 Live, on November 10, 2026, according to the company's announcement. The new speech-to-text models cut word and diarization errors sharply compared with the previous Scribe v2 lineup across 102 languages, while keeping per-hour pricing unchanged.

What changed

• Scribe v3, built for batch and recorded audio, makes 27% fewer word errors than Scribe v2 across 102 languages on the FLEURS benchmark, plus 17% fewer diarization errors on multi-speaker recordings.
• Scribe v3 Live, built for realtime audio, makes 43% fewer errors than Scribe v2 Realtime.
• Scribe v3 Live adds ElevenLabs' first realtime speaker diarization, with 39% fewer diarization errors than the competitor models it cites.
• Scribe v3 Live adds a multilingual mode that handles mid-sentence language switching and uses full conversational context for turn detection.

What it means for people who build with AI

Scribe is widely embedded in voice agents, dictation tools and coding assistants, so more accurate transcription and realtime diarization at the same price flow directly into every product built on the API, without forcing developers to change how they pay for it.

Availability and pricing

• Pricing stays unchanged: Scribe v3 costs $0.20 per hour of audio and Scribe v3 Live costs $0.22 per hour, the same as Scribe v2 and v2 Realtime. Keyterm prompting adds $0.05 per hour and transcript editing adds $0.07 per hour.
• Both models are now the default speech-to-text models in the ElevenLabs app; Scribe v3 Live is also available through ElevenAgents and ElevenAPI.
• Deployment regions include the US, EU, India and Singapore, and a Zero Retention Mode deletes audio and text immediately after a request completes.

Sources

• Introducing Scribe v3 and v3 Live, our most accurate speech to text models (ElevenLabs, November 10, 2026, official)

Written with AI assistance, strictly from the sources listed above. Before it went live, automated checks verified its dates, sources and every language edition.