Eleven v4
Eleven v4 is ElevenLabs' flagship text-to-speech model, launched on 28 September 2026 on a new architecture. It reads scripts with awareness of tone, pacing and context, supports over 90 languages and multi-speaker dialogue, and is available in ElevenLabs apps and…
Best for
High-quality, expressive narration, dubbing, audiobooks and multi-speaker dialogue where quality matters more than latency.
Overview
Eleven v4 is ElevenLabs' flagship text-to-speech model, launched on 28 September 2026 on a new architecture. It reads scripts with awareness of tone, pacing and context, supports over 90 languages and multi-speaker dialogue, and is available in ElevenLabs apps and the API.
Strengths and limits
- Most expressive ElevenLabs model, steerable with inline tags and natural-language direction
- More than 90 languages, multi-speaker dialogue, cross-language voice identity
- Instant Voice Clone from 10 seconds of audio; Professional Voice Clones supported
- Closed model, API and app access only
- Higher price per character than v4 Turbo; not the latency-optimised option for live agents
Published benchmark scores
Scores are as published by the source; settings (reasoning effort, tools, attempts) differ between labs, so compare with care. How to read benchmarks →