Eleven v4 and the ElevenLabs models
ElevenLabs is the company most people mean by AI voice. Eleven v4, released on 28 September 2026, is its expressive flagship, and Eleven v4 Turbo the low-latency version of it for agents; Eleven v3, the previous flagship, keeps its page. Flash and Turbo v2.5 are the older low-latency lines, and Multilingual v2 the long-standing standard. Each page with a model reads the same four scripts as every voice model on this site.
ElevenLabs · facts checked
Which one to pick
Eleven v4
$0.08 per 1,000 characters via ElevenLabs API
Expressive narration and voiceover; it returned faster than v3 for us.
Eleven v4 Turbo
$0.04 per 1,000 characters via ElevenLabs API
Real-time agents over ElevenLabs' streaming API, or v4 at half the price.
Eleven v3
$0.08 per 1,000 characters via ElevenLabs API
A pipeline already tuned to v3's delivery.
Side by side, measured
| Model | Price | Seconds to return | Cost of its fixed set | Voices |
|---|---|---|---|---|
| Eleven v4 | $0.08 per 1,000 characters via ElevenLabs API | 4 to 8 | $0.078 4 scripts | Twenty named stock voices on the hosted endpoint we used, the same list as Eleven v3; we used Sarah, as on v3 |
| Eleven v4 Turbo | $0.04 per 1,000 characters via ElevenLabs API | 4 to 6 | $0.039 4 scripts | The same twenty named stock voices as Eleven v4 on the hosted endpoint we used; we used Sarah |
| Eleven v3 | $0.08 per 1,000 characters via ElevenLabs API | 7 to 10 | $0.098 4 scripts | Twenty named stock voices on the hosted endpoint we used, from Aria to Bill; we used Sarah |
Seconds and cost are from our own runs of each model's fixed set: the same 4 scripts for every model that makes readings from text. Open a model for every reading and the request behind it.
Every version
| Version | Released | What changed |
|---|---|---|
| Multilingual v2 | The long-standing default, 29 languages. | |
| Turbo v2.5 | Low latency for conversational use. | |
| Flash v2.5 | About 75 ms latency for agents. | |
| Eleven v3 (alpha) | Audio tags, dialogue mode, 70-plus languages. | |
| Eleven v3 eleven_v3 | General availability, fewer text errors, on the API. | |
| Eleven v4 eleven_v4 | New architecture, more than 90 languages, natural-language direction, $0.08 per thousand characters. | |
| Eleven v4 Turbo | Low-latency v4 for agents, about 150 ms to first speech by ElevenLabs' measure, $0.04. |
Frequently asked questions
Is ElevenLabs free?
A free plan gives 10,000 credits a month for non-commercial use. Paid plans start at $6.
Which ElevenLabs model should I use?
Eleven v4 for finished voiceover and v4 Turbo for real-time agents or a lower price, both released on 28 September 2026. On ElevenLabs' API v4 is $0.08 per thousand characters and Turbo $0.04, read on 22 October 2026.