SonnaLabs gives you five ElevenLabs models in the same studio, next to HeyGen and Google Cloud voices. They differ in how expressive they are, how many languages they speak and how fast they answer. This guide puts them side by side.
The short version
- Eleven v4 is the most emotive model, with support for more than 90 languages. Choose it when the performance matters most: narration, characters, anything that has to sound alive.
- Eleven v4 Turbo is the same family tuned for low latency, also with more than 90 languages, for fewer credits than Eleven v4. Choose it for quick drafts, long scripts and anything you will generate many times.
- Eleven v3 is the most expressive earlier generation, with more than 70 languages and a wide emotional range.
- Multilingual v2 gives life-like, emotionally rich voices in 29 languages and is built for professional audio.
- Flash v2.5 is the ultra-low latency model for responsive, real-time uses, in 32 languages, priced like Eleven v4 Turbo.
What each one costs
Text to speech is priced per 1,000 characters. The table below is read from the SonnaLabs catalog, so it always shows the current prices.
| Model | Configuration | Price in credits |
|---|---|---|
 Eleven v4ElevenLabs | Standard | 2,100 / 1K characters $0.147 |
| Eleven v4 TurboElevenLabs | Standard | 1,050 / 1K characters $0.073 |
| Eleven v3ElevenLabs | Standard | 2,100 / 1K characters $0.147 |
| Multilingual v2ElevenLabs | Standard | 2,100 / 1K characters $0.147 |
| Flash v2.5ElevenLabs | Standard | 1,050 / 1K characters $0.073 |
Prices are in credits; the dollar amounts are estimates. The current prices are always on the models page.
How to choose
- Start with Eleven v4 Turbo when you are still writing. It costs fewer credits than Eleven v4 and answers faster, so you can listen to a script, change it and listen again.
- Switch to Eleven v4 for the final take when the voice has to carry feeling.
- Use Multilingual v2 or Eleven v3 when you want the sound of those generations, or when your language is covered by them.
- Use Flash v2.5 when speed matters more than anything else.
Voices and languages
The model decides how a voice performs; the voice decides who is speaking. On the text to speech page you can browse the voices, filter them by language, category, gender and age, and play a sample before you commit credits.
ElevenLabs models are part of the paid plans and credit packs. The free plan covers Google Cloud and HeyGen Voice Turbo voices; see how credits work.