Reported News
ElevenLabs launches Scribe v3 and Scribe v3 Live speech-to-text models
ElevenLabs launched Scribe v3 and Scribe v3 Live on Nov 10, 2026, calling them "our most accurate speech to text models." The company says Scribe v3 makes 27% fewer word errors than Scribe v2 across 102 languages, and Scribe v3 Live 43% fewer errors than Scribe v2 Realtime.
- Changes reported
- 6
- listed below, each from a source we read
- We tested
- Not tested yet
- reported from sources; we did not run it ourselves
- Gaps we flag
- 4
- things we could not verify
- Sources
- 1
- 1 primary, linked below
What changed
| # | What the source says | Status |
|---|---|---|
| 01 | Scribe v3 is the new model for recorded audio. ElevenLabs says it makes 27% fewer word errors than Scribe v2 across 102 languages on an internal version of the FLEURS benchmark, improving in every language it tested. | Reported |
| 02 | Scribe v3 Live is the low-latency realtime variant. The post says it cuts semantic word error rate by 43% and WER by 31% versus Scribe v2 Realtime, and adds speaker diarization to realtime transcription for the first time. | Reported |
| 03 | Turn detection in Scribe v3 Live uses conversation context as well as audio, so a caller who pauses mid-number is not cut off. ElevenLabs says this is the turn detection ElevenAgents already used, now exposed in ElevenAPI. | Reported |
| 04 | Keyterm prompting takes up to 1,000 terms, transcript edits apply formatting rules written in plain English, and numbers can return as digits or words. Scribe v3 Live adds a multilingual mode for mid-sentence language switching and a keyterm library for larger vocabularies. | Reported |
| 05 | On diarization, the post reports 12% fewer errors than Scribe v2 overall and 17% fewer on recordings with three or more speakers; Scribe v3 Live scored a 3.1% diarization error rate on English real-world audio versus 5.9% to 7.3% for other realtime models tested. | Reported |
| 06 | Both models are in ElevenAPI today and are the default in the ElevenLabs app; existing integrations switch by changing the model ID to scribe_v3 or scribe_v3_live. They run in the US by default, with isolated EU, India and Singapore environments for enterprise data residency. | Reported |
reported by the sourcemeasured or run by usnot verified
Primary source: ElevenLabs blog: Introducing Scribe v3 and v3 Live. We are the fast, independent index; the vendor’s own page is the source of record for this release.
In the source’s words
“In languages easier to transcribe, German improved by 42%, Spanish by 35%, and English by 23%.”
What we have not verified
- We did not run Scribe v3 or Scribe v3 Live; every accuracy, WER and diarization figure is from ElevenLabs' own evaluations, including an internal version of FLEURS and an in-house dataset that is not yet public.
- The post does not state per-minute or per-hour pricing for either model.
- The customer result (40% more Dutch postcodes recognized on first try) is a vendor-reported pre-launch test that we could not check.
- ElevenLabs says it will open source its 16-language real-world dataset soon; no date or link was given.
This brief is Reported: written from the sources below, not from running the release ourselves.
Sources
- ElevenLabs blog: Introducing Scribe v3 and v3 Liveprimary source · fetched 2026-11-11
Related
Spotted an error? See our corrections policy or email the address on the contact page. Markdown version for agents: /news/elevenlabs-scribe-v3-speech-to-text.md
Want these briefs in your Google Top Stories? Add Agentic Weekly as a preferred source on Google.
Newsletter
The Weekly Verdict. Free, every Monday.
A free email every Monday: the week's few big AI releases, then 5 short use cases for the new models, each with a prompt you can paste.
Free. Double opt-in. One click to unsubscribe.