ElevenLabs Debuts Eleven v4, Tops Voice TTS Rankings With Expressive Architecture

ElevenLabs launched Eleven v4 and Eleven v4 Turbo on September 28, 2026, marking a shift in text-to-speech expressiveness through a redesigned architecture that interprets tone, pacing, emotion, and character from written text. The company says v4 became the first speech synthesis model to rank first on the Artificial Analysis Provider Voice TTS Arena, with an Elo rating of 1319 and an 91.7 percent pronunciation accuracy score—up from 85.6 percent in v3.