HaiAI123

Curated Global AI Tools Directory

Industry update
Flat

Zoom Scribe V2.1 Pro Live tops AA-WER streaming transcription leaderboard

11/20/2026 — 11/20, 08:18·1 sources (all)·3 reports (all)

Key takeaways

  • Artificial Analysis independently tested Zoom's new streaming speech-to-text model, Scribe V2.1 Pro Live. It ranked first on the AA-WER Stre…

Background & analysis

On 2026-11-20 at 07:40, the benchmarking firm Artificial Analysis published the results of its independent tests of Zoom's new streaming speech-to-text model, Scribe V2.1 Pro Live, through its X account. The firm reported that Scribe V2.1 Pro Live ranked first for final transcript accuracy on AA-WER Streaming, posting a 2.4% word error rate, with MAI-Transcribe-2-Streaming close behind at 2.5%. Artificial Analysis also offered congratulations to the Zoom team.

Further results released the same day clarified the standings within the 40 models covered by the AA-WER Streaming leaderboard. Scribe V2.1 Pro Live took the top spot with a 2.4% final transcript WER, edging out MAI-Transcribe-2-Streaming at 2.5%. On latency, however, it was the slower of the two: the transcript was returned 0.66s after end of speech, against 0.13s for MAI-Transcribe-2-Streaming. At the first partial, or interim, transcript issued 0.55s after end of speech, Scribe V2.1 Pro Live recorded a 5.0% WER, whereas MAI-Transcribe-2-Streaming reached 2.5% in just 0.12s, making its first interim result both faster and more accurate.

No response from Zoom has appeared in the material, so the public record currently rests on the test figures released by Artificial Analysis on that date.

AI-generated from 3 reports · updated 2 days ago

Latest turnIndependent testing by Artificial Analysis finds that Zoom's new streaming speech-to-text model, Scribe V2.1 Pro Live, scores 5.0% WER on its first partial transcript, delivered 0.55s after end of speech. On that measure MAI-Transcribe-2-Streaming is both more accurate and faster, at 2.5% and 0.12s.

updream
24-hour heatHeat index 25 · peak 45 · 23h ago
24 hours agonow

Reports on this story headlines open the original

Nov 20
  1. Independent testing by Artificial Analysis finds that Zoom's new streaming speech-to-text model, Scribe V2.1 Pro Live, scores 5.0% WER on its first partial transcript, delivered 0.55s after end of speech. On that measure MAI-Transcribe-2-Streaming is both more accurate and faster, at 2.5% and 0.12s.

    X:Artificial Analysis (@artificialanlys)First-partyAI score 41
  2. Zoom Scribe V2.1 Pro Live ranks first among 40 models on Artificial Analysis' AA-WER Streaming benchmark with 2.4% final transcript WER, just ahead of MAI-Transcribe-2-Streaming at 2.5%. It returns that transcript more slowly, though: 0.66s after end of speech versus 0.13s.

    X:Artificial Analysis (@artificialanlys)First-partyAI score 49
  3. Artificial Analysis says it independently benchmarked Zoom's new streaming speech-to-text model, with Scribe V2.1 Pro Live taking the top spot for final transcript accuracy on AA-WER Streaming at 2.4% WER, just ahead of MAI-Transcribe-2-Streaming at 2.5%. The firm congratulated the Zoom team.

    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 →