Release·Feb 2, 2026
Eleven v3
The most expressive TTS model hits GA, with audio tags and multi-speaker dialogue.
11
ElevenLabsEleven v3, ElevenLabs' most expressive text-to-speech model, graduated from alpha to general availability across all platforms. It spans 70+ languages and adds inline “audio tags” like [whispers] or [laughs] plus a Text-to-Dialogue API for multi-speaker, emotionally reactive speech — well suited to game cinematics and pre-rendered character VO (ElevenLabs still points truly real-time agents to its faster v2.5 model).
ElevenLabsVoice
11
ElevenLabs
Best for placeholder or shipped voice work when you can't hire a full VO cast.