> ## Documentation Index
> Fetch the complete documentation index at: https://docs.kymaapi.com/llms.txt
> Use this file to discover all available pages before exploring further.

# ElevenLabs v3

> ElevenLabs v3 is a text-to-speech model optimized for high-fidelity, emotionally expressive audio generation. Reach for it when voice quality, pacing control, and multilingual coverage matter more than raw throughput or latency.

## Overview

ElevenLabs v3 generates speech from text input with support for emotional range, inline audio tags, and natural delivery across more than 70 languages. It operates at a medium speed tier with a balanced cost profile and a 5,000-token context window.

On Kyma, the model runs through an OpenAI-compatible endpoint with automatic failover to maintain delivery if a serving path degrades. It supports prompt caching, which bills repeated prompt prefixes at this model's cached input rate. Responses return exact billing data in usage.cost and identify the executed model via the X-Kyma-Model header.

The model does not support reasoning, vision, or structured outputs. It is strictly a speech generation engine, so it will not process images, execute code, or return JSON schemas.

## Specs

| Field              | Value                                                                          |
| ------------------ | ------------------------------------------------------------------------------ |
| Model ID           | `eleven-v3`                                                                    |
| Best for           | Expressive narration, character voices, emotional dialogue, premium voiceovers |
| Context window     | 5,000                                                                          |
| Input modalities   | Text                                                                           |
| Output modalities  | Audio                                                                          |
| Tool calling       | Yes                                                                            |
| Structured outputs | Yes                                                                            |
| Prompt caching     | Yes                                                                            |
| Speed              | Medium                                                                         |
| Cost band          | Balanced                                                                       |
| Release stage      | Stable                                                                         |

## Pricing

|        | Per 1M tokens |
| ------ | :-----------: |
| Input  |     \$0.20    |
| Output |     \$0.60    |

## Use this when

* **Premium Voiceover Production** — Generate broadcast-quality narration with controlled emotional pacing and tone.
* **Character Dialogue Generation** — Produce distinct, expressive voices for interactive fiction or game assets.
* **Multilingual Audio Localization** — Convert scripts into natural-sounding speech across more than seventy languages.
* **Emotional Audiobook Narration** — Render long-form text with dynamic inflection and audio tag directives.

## Not ideal for

Do not use this model for real-time conversational speech or high-throughput batch processing where latency and cost efficiency are the primary constraints.

## Pick something else when

* You need faster, lower-latency speech generation: use [`eleven-flash-v2-5`](/models/eleven-flash-v2-5).
* You require custom voice cloning or voice design: use [`minimax-voice-clone`](/models/minimax-voice-clone) or [`minimax-voice-design`](/models/minimax-voice-design).
* You need background music or sound effects instead of speech: use [`elevenlabs-music`](/models/elevenlabs-music) or [`elevenlabs-sfx`](/models/elevenlabs-sfx).

## Example

```python theme={null}
from openai import OpenAI

client = OpenAI(base_url="https://kymaapi.com/v1", api_key="ky-...")

response = client.chat.completions.create(
    model="eleven-v3",
    messages=[{"role": "user", "content": "..."}],
)
```

## Agent query example

Ask the API which models fit, instead of hardcoding an id:

```bash theme={null}
curl "https://kymaapi.com/v1/models?recommended_for=generate-speech&tools=true&quality_tier=frontier-open"
```

## FAQ

**Does this model support prompt caching on Kyma?**

Yes. Repeated prompt prefixes are billed at this model's cached input rate, reducing costs for recurring scripts.

**What happens if the serving path degrades during a request?**

Kyma automatically reroutes the request to a healthy provider, ensuring delivery without manual intervention.

**Can I use this model for structured JSON outputs or image analysis?**

No. It only accepts text input and returns audio. It does not support reasoning, vision, or structured output formats.