Loading…
Get the best builds by email. No account needed.
Live drop
ElevenLabsDropped Nov 10, 2026
43% fewer errors than prior realtime model with first-in-class diarization and 150ms latency.
About this release
Scribe v3 Live is ElevenLabs' low-latency realtime speech-to-text model. It cuts semantic word error rate by 43% and WER by 31% versus Scribe v2 Realtime across 73+ languages, adds speaker diarization, improves turn detection using full conversation context, and supports multilingual switching plus keyterm libraries. Available in ElevenAgents and via ElevenAPI.
Read the official ElevenLabs launch pageFirst builds incoming — this lane fills in automatically as people start building with Scribe v3 Live. Know one? Submit the URL.