ElevenLabs' new v4 speech model makes AI voices more expressive and consistent
ElevenLabs released Eleven v4 speech model with better audio-tag adherence, 90+ languages, and a 150ms Turbo variant for real-time voice agents.
ElevenLabs launched Eleven v4, a speech model with a new architecture that analyzes script tone, pacing, and context, supports more than 90 languages, and handles up to 10,000 characters per request. It scores 91.7 percent on pronunciation benchmarks, up from v3's 85.6, and ranks ahead of Cartesia Sonic 3.6 and Google's Gemini 3.8 Flash TTS on Artificial Analysis' Provider Voice Arena. The v4 Turbo variant starts producing speech in about 150 milliseconds, versus 262ms for Cartesia Sonic 3.6 and 814ms for OpenAI's GPT-4o mini TTS. Standard API pricing is $80 per million characters ($40 for Turbo), temporarily cut to $22 and $11 through October 12.
- New architecture improves audio-tag adherence and cross-segment voice consistency
- Turbo variant starts producing speech in roughly 150 milliseconds
- Supports 90+ languages with 91.7% pronunciation benchmark score
- Temporary price cuts run through October 12
- Organizations
- ElevenLabsCartesiaGoogleOpenAI
From the original article
3 min read at the-decoder.com
Elevenlabs' new speech model, Eleven v4, follows cues for laughter and whispering more accurately and keeps voices consistent across long productions like audiobooks. Its Turbo variant starts speaking in 150 milliseconds and is built for real-time voice agents. On Artificial Analysis' Voice Arena leaderboard, v4 ranks ahead of Cartesia and Google's Gemini.
By Jonathan Kemper · the-decoder.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.