ElevenLabs releases Eleven v4 and v4 Turbo TTS models

On September 28, 2026, ElevenLabs released the eleven_v4 TTS model for content via the Text to Dialogue API and eleven_v4_turbo for voice agents via WebSocket — featuring 90+ languages, inline audio tags instead of SSML, restored Professional Voice Clones, and up to 10,000 characters per generation.

Compiled by Sergey KostenchukPublished 2026-10-05Updated 2026-10-06

2026-10-05 Multimodal OpenAI

🎨 ElevenLabs releases Eleven v4 and v4 Turbo TTS models

On September 28, eleven_v4 for content via the Text to Dialogue API and eleven_v4_turbo for voice agents via WebSocket were released. 90+ languages, with tags like [laughs], [whispers], [pause] replacing the disabled SSML. Professional Voice Clones have been restored, but old clones need retraining. Up to 10,000 characters with "context stitching."

🌍 v4 Turbo is claimed to have ~150 ms time to first speech, compared to 262 ms for Cartesia Sonic 3.6 and 814 ms for GPT-4o mini TTS — a claim for the realtime segment of voice agents. You need to change the model_id and retrain clones to avoid being left on Multilingual v2.

👤 Claude Opus 5.5 wrote the code that allows Reachy Mini to express voice emotions outwardly — through antenna movement. Until October 12, Creator+ plans offer 3x credits: Russian can be voiced among 90+ languages more cheaply. Cloning someone else's voice requires legal review.

Source 1: https://elevenlabs.io/v4 Source 2: https://elevenlabs.io/docs/changelog

Sources