On September 28, 2026, @elevenlabsio launched Eleven v4, which it calls its most emotive text-to-speech model yet, together with a low-latency sibling, Eleven v4 Turbo. Want to try it right away? Start here: https://try.elevenlabs.io/vrwplh9l9my0 (affiliate link).
How this post was put together: every fact below comes from official ElevenLabs sources: the launch blog post, the Eleven v4 product page, and the ElevenLabs docs, checked October 26, 2026. This is a summary, not a hands-on test, and we don't work for ElevenLabs.
What's new in Eleven v4
- A new architecture. ElevenLabs says v4 is built on an entirely new architecture that reads a script "the way a voice actor would", interpreting tone, pacing, emotion, character and context.
- Direction inside the script. Add inline audio tags like
[laughs],[whispers],[light rain]or[door slams], or describe delivery in natural language. According to ElevenLabs, v4 follows tag sequences and direction prompts more reliably than v3, sound effects included. - Natural multi-speaker dialogue. Because the model understands the context of a whole scene, speakers respond to what was just said instead of sounding like stitched-together lines.
- Consistent voices. A new method for capturing speaker identity keeps a voice stable across regenerations, dialogue and narration.
- Long-form without seams. Context stitching keeps pacing and delivery steady across long scripts such as audiobooks. A single request handles up to 10,000 characters (about 10 minutes of audio).
- 90+ languages.
- Professional Voice Clones are back. They weren't supported in v3; in v4 they work with the model's full emotional range.
- Better pronunciation control. Improved IPA phoneme support, plus Pronunciation Dictionaries for names, acronyms and technical terms.
Eleven v4 vs Eleven v4 Turbo: which one to pick
| Eleven v4 | Eleven v4 Turbo | |
|---|---|---|
| Model ID | eleven_v4 |
eleven_v4_turbo |
| Best for | Expressive or long-form narration, audiobooks, character voiceovers | Real-time experiences and voice agents |
| Speed | — | Median inference latency ~100 ms, time to first speech ~150 ms (per ElevenLabs) |
| Audio tags | Yes | Yes |
| Where | ElevenLabs app and API | API and ElevenAgents |
Who Eleven v4 is for
- Audiobook and long-form narrators who need consistent delivery from first page to last.
- Game, animation and fiction creators writing characters and multi-speaker scenes.
- YouTubers and ad producers who want emotional range and built-in sound effects in voiceovers.
- Developers building voice agents (via Eleven v4 Turbo and ElevenAgents) who need expressive speech in real time.
- Multilingual teams producing content across many languages.
How to try Eleven v4
- Create an ElevenLabs account at https://try.elevenlabs.io/vrwplh9l9my0.
- In the app, open Text to Speech, choose Eleven v4 as the model, pick a voice and paste your script.
- Add direction with audio tags, for example:
[warm] Welcome back. [long pause] [whispers] Here's the secret. - Via the API, pass
model_id: "eleven_v4"(or"eleven_v4_turbo"for real time) to the Create speech or Stream speech endpoints, or use the Create dialogue endpoint for multiple speakers. ElevenLabs offers TypeScript and Python SDKs.
See the official audio tags guide for the full tag syntax.
Sources
- Introducing Eleven v4, our most emotive model, ElevenLabs blog (published Sep 28, 2026)
- Eleven v4 and Eleven v4 Turbo, ElevenLabs
- ElevenLabs Documentation: Which model should I use?
- Text to Speech capabilities
Frequently asked questions
What is Eleven v4?
+
Eleven v4 is the text-to-speech model ElevenLabs describes as its most emotive and expressive yet. It launched on September 28, 2026 alongside a low-latency variant, Eleven v4 Turbo.
What is the difference between Eleven v4 and Eleven v4 Turbo?
+
Eleven v4 (model_id eleven_v4) is recommended for expressive or long-form narration. Eleven v4 Turbo (model_id eleven_v4_turbo) targets real-time use and voice agents, with a median inference latency of about 100 ms according to ElevenLabs.
How many languages does Eleven v4 support?
+
ElevenLabs lists 90+ languages for Eleven v4, and says Eleven v4 Turbo covers the same languages in real time.
Does Eleven v4 support audio tags?
+
Yes. You can add inline tags such as [laughs], [whispers] or [door slams] to the script. The same tags work with Eleven v3 and v4, and v4 Turbo also supports them.
Can I use a Professional Voice Clone with Eleven v4?
+
Yes. ElevenLabs notes that Professional Voice Clones weren't supported in v3 but are back in Eleven v4.
