HaiAI123

Curated Global AI Tools Directory

Industry update
Flat

ElevenLabs Scribe v3 ranks second on AA-WER non-streaming benchmark

11/20/2026 — 11/20, 10:21·1 sources (all)·1 reports (all)

Key takeaways

  • Artificial Analysis' independent testing puts ElevenLabs' Scribe v3 second on its non-streaming AA-WER speech-to-text index at 1.95% word er…

Background & analysis

On November 20, 2026, the evaluation firm Artificial Analysis published independent benchmark results on its X account, revealing that ElevenLabs' Scribe v3 ranked second on the AA-WER Index for non-streaming speech-to-text. It achieved a word error rate of 1.95%, a clear improvement over its predecessor Scribe v2, which recorded a word error rate of 2.18%. In the same benchmark, Scribe v3 trailed Alibaba's Fun-Realtime-ASR-preview, which posted a 1.73% word error rate, while holding only a slim lead over Microsoft AI's MAI-Transcribe-2, which came in at 2.04%. These figures place Scribe v3 in the runner-up position for non-streaming speech-to-text, between the two competing systems, with the margin over MAI-Transcribe-2 particularly narrow. The published benchmark results stand as the latest available update on this ranking, and they fix the current standings and word error rate measurements for the models covered on the AA-WER Index for non-streaming speech-to-text. Artificial Analysis presented the data as its own independent testing, with no further rankings, revised figures, or additional product entries mentioned at the time of publication.

AI-generated from 1 reports · updated 2 days ago

Related AI tools

updream
24-hour heatHeat index 25 · peak 47 · 23h ago
24 hours agonow

Reports on this story headlines open the original

Nov 20
  1. ElevenLabs' Scribe v3 ranks second on Artificial Analysis's AA-WER Index for non-streaming speech-to-text at 1.95% word error rate, improving on Scribe v2's 2.18%. It trails Alibaba's Fun-Realtime-ASR-preview at 1.73% and sits just ahead of Microsoft AI's MAI-Transcribe-2 at 2.04%.

    X:Artificial Analysis (@artificialanlys)First-partyAI score 46

Other stories people are talking about

How is heat calculated?About the method

Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.

This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.

Surge
Discussion rising fast
New
First report within 6 hours
Rising
Still gathering discussion

Back to AI industry updates →