Eleven v4 is ElevenLabs' newer text-to-speech generation, while Eleven v3 remains available in MakeFX for existing recipes and for voices or performances you prefer. Choose by listening to the same short passage in both models; v4's stronger source-voice matching can make a cloned voice sound noticeably different from its v3 result.
| Capability | Eleven v4 | Eleven v3 |
|---|---|---|
| Provider model | eleven_v4 | eleven_v3 |
| Character limit | 10,000 | 5,000 |
| Delivery direction | Inline audio tags, including emotion and reactions | Expressive audio tags |
| Voice controls in MakeFX | Stability and Similarity | Stability and Speed |
| Current configured standard rate | $0.08 / 1,000 characters | Existing v3 catalog rate |
The provider limits v4 to 10,000 characters and v3 to 5,000; MakeFX's request contract caps a prompt at 8,000 characters for every model.
Eleven v4 Turbo is a third choice for v4-style speech generation with a lower-latency focus and a $0.04 per 1,000-character standard reference rate. These provider reference rates exclude the MakeFX managed markup and minimum fee. Your exact credit quote appears before generation and depends on prompt length and billing mode. The provider's temporary v4 launch discount is not included in MakeFX's standard reference rates.
For multilingual use, Eleven v4 aims for a natural target-language accent when the output language differs from the voice's source language. If preserving a particular cross-language accent or an existing v3 sound matters, compare a short sample first. See the Eleven v4 model guide and the ElevenLabs API price list for provider details.
Choose Eleven v4 for speech generation, or follow the audio playbook to create speech assets with reusable prompts and inspectable word timings.