Oakgen.aiElevenLabs Eleven v4 Expressive AI Voice with Audio Tags
- Type
- Audio
- Made by
- ElevenLabs
- Version
- v4
- Released
- 2026-09-28
Specs
- Max resolution
- MP3 audio
- Inputs → outputs
- text → audio
Compare
| Model | Starts at · credits |
|---|---|
| ElevenLabs Eleven v4This pageElevenLabs | — |
| ElevenLabs v3ElevenLabs | — |
| ElevenLabsElevenLabs | — |
| MiniMax Speech HDMiniMax |
What is ElevenLabs Eleven v4?
ElevenLabs' newest expressive voice model, with audio tags in 90+ languages.
Eleven v4 is ElevenLabs' newest text-to-speech model, released on September 28, 2026, together with Eleven v4 Turbo for real-time use. Both speak 90+ languages, and v4 takes up to 10,000 characters per request. You steer delivery with audio tags such as [whispers] or [laughs]. On Oakgen, V4 is the default ElevenLabs model.
Capabilities at a glance
- 90+ languages in both v4 and v4 Turbo
- Up to 10,000 characters per request on v4
- Audio tags in square brackets, such as [whispers], [laughs], [sighs] and [excited]
- Eleven v4 Turbo for real-time use, about 100 ms median inference according to ElevenLabs
- Pronunciation control with IPA wrapped in forward slashes
- Stability and Similarity settings on Oakgen
How to use ElevenLabs Eleven v4
Write the script for the ear
Paste up to 10,000 characters. Write the words a speaker would actually say, in short sentences, and pick a voice from the ElevenLabs library.
Mark each change in delivery with an audio tag
Put a tag in square brackets before the line it shapes, such as [whispers], [excited], [sighs] or [laughs]. ElevenLabs says v4 follows tags more reliably than earlier models.
Add the next tag where the mood shifts
A tag carries forward until the next one. If a whisper should end after one line, start the following line with a new tag rather than leaving the rest of the script whispered.
Fix pauses and pronunciation
v4 does not support SSML <break> tags, so shape pauses with audio tags and the wording itself. For a name or brand word, wrap its IPA in forward slashes, such as /ˈoʊkdʒɛn/.
Choose V4 or V4 Turbo and generate
Pick V4 for the most expressive read or V4 Turbo for faster, lower-priced speech. Set Stability and Similarity, generate, and listen back before you export.
Compared to other models
- vs. elevenlabs-v3 — languages and script length
- Eleven v4 covers 90+ languages and up to 10,000 characters per request, against 70+ languages and 5,000 characters for v3. Both use audio tags. On Oakgen, v3 keeps the Style and Speed settings that v4 does not expose.
License & commercial use
Licensed through ElevenLabs' commercial terms.
Permitted on paid Oakgen plans subject to ElevenLabs terms. Clone a voice only with the speaker's consent.
Frequently asked questions
6What is the difference between Eleven v4 and Eleven v4 Turbo?
ElevenLabs released both on September 28, 2026, and both speak 90+ languages. v4 is the expressive model. v4 Turbo targets real-time use, with a median inference time of about 100 ms according to ElevenLabs, and costs half as much per character.
How much does Eleven v4 cost?
ElevenLabs' API list price in October 2026 is $0.08 per 1,000 characters for v4 and $0.04 per 1,000 characters for v4 Turbo. On Oakgen you pay in credits per character, worked out from that provider cost plus Oakgen's platform fee.
How do audio tags work in Eleven v4?
Write a tag in square brackets, such as [whispers], [laughs], [sighs] or [excited], before the text it should shape. The tag carries forward until the next tag. ElevenLabs says v4 follows tags more reliably than earlier models.
Does Eleven v4 support SSML break tags?
No. ElevenLabs does not support SSML <break> tags on v3 or v4. Shape pauses with audio tags and the wording of the script, and wrap IPA in forward slashes to fix how a word is pronounced.
Where can I use Eleven v4 on Oakgen?
In the voice generator, in Agent Chat and through Oakgen's MCP server. V4 is the default ElevenLabs model, next to V4 Turbo, V3 and Multilingual V2. On V4 and V4 Turbo, Oakgen exposes the Stability and Similarity settings.
Can I use Eleven v4 audio commercially?
Yes, on paid Oakgen plans subject to ElevenLabs terms. Clone a voice only with the speaker's consent.