kapynAI / Models

ElevenLabs' new v4 speech model makes AI voices more expressive and consistent

ElevenLabs launches Eleven v4, a speech model that follows laughter cues and maintains voice consistency. The Turbo variant can speak in 150 ms, enabling real‑time voice agents, and it tops the Voice Arena leaderboard ahead of Cartesia and Google’s Gemini. This improvement matters for developers building long‑form audio or interactive voice applications.

The Decoder·Sep 29, 2026

Opening Kapyn…