Technical specifications of Eleven v4
| Item | Specification |
|---|---|
| Model name | Eleven v4 |
| CometAPI Model ID | eleven_v4 |
| Original provider | ElevenLabs |
| Model category | Text-to-Speech (TTS) |
| Primary API format | Runway-compatible Text-to-Speech API |
| CometAPI endpoint | POST /runwayml/v1/text_to_speech |
| Input | Text prompt and preset voice configuration |
| Output | Generated speech audio; CometAPI announces MP3 output |
| CometAPI text limit | Up to 2,500 characters per request, according to its October 10, 2026 announcement |
| Native ElevenLabs text limit | 10,000 characters per request |
| Language support | 90+ languages in the native Eleven v4 model |
| Voice controls | Stability, similarity boost, style, speed and speaker boost, subject to gateway support |
| Expressive controls | Tags such as [laughs], [whispers] and [pause] |
| Multi-speaker dialogue | Supported by the native Eleven v4 model through the Text to Dialogue API |
| Processing | Asynchronous task submission and status polling through CometAPI |
| Audio format | MP3 output announced by CometAPI; confirm available format options in the current endpoint schema |
Sources: CometAPI model announcement and CometAPI Runway Text-to-Speech API documentation. Native model capabilities are documented by ElevenLabs.
What is Eleven v4?
Eleven v4 is ElevenLabs' expressive speech-synthesis model, designed to generate natural-sounding speech with rich emotional delivery, contextual awareness, and strong voice consistency. It is particularly suitable for narration, character voiceovers, audiobooks, and dialogue where the way a line is performed matters as much as the words themselves.
Through CometAPI, the model is available using the identifier eleven_v4 via a Runway-compatible text-to-speech interface. The gateway adds its own request format and constraints, so CometAPI's documented character limit should be used when integrating the model rather than assuming the native ElevenLabs API limit applies.
Main features of Eleven v4
- Expressive speech synthesis: Produces speech with nuanced emotion, emphasis, pacing and delivery, making it suitable for expressive scripts rather than flat, uniform narration.
- Natural voice performance: Generates human-like speech for characters, storytelling, professional narration and conversational dialogue.
- Multilingual generation: The native model supports more than 90 languages, including English, Japanese, Mandarin Chinese, Korean, French and Spanish.
- Fine-grained voice control: Supported parameters include stability, similarity boost, style, speed and speaker boost, allowing developers to tune delivery and voice consistency.
- Emotion and delivery tags: Tags such as
[laughs],[whispers]and[pause]can guide expressive delivery in supported requests. - Asynchronous API integration: CometAPI accepts a speech-generation task and returns a task ID, which can be used to retrieve status and the resulting audio.
Feature availability and input limits may differ between native ElevenLabs endpoints and the CometAPI gateway.
Eleven v4 vs. Eleven v4 Turbo vs. Eleven v3
| Model | Main strength | Best use case |
|---|---|---|
| Eleven v4 (eleven_v4) | Rich emotional expression and high-fidelity speech | Audiobooks, narration, animation and character dialogue |
| Eleven v4 Turbo (eleven_v4_turbo) | Expressive speech optimized for low latency; native median inference latency around 100 ms | Interactive voice applications and real-time agents |
| Eleven v3 (eleven_v3) | Expressive speech with dramatic delivery and audio-tag control | Emotion-heavy voice performances and character scenes |
| Eleven Multilingual v2 (eleven_multilingual_v2) | Stable multilingual speech quality, particularly for long-form content | Consistent narration and multilingual production |
These comparisons describe the native ElevenLabs models. Availability and performance through CometAPI depend on its exposed endpoint and implementation.
Representative use cases
- Audiobook production: Generate expressive narration with natural pacing and character differentiation.
- Animation and gaming: Create character dialogue with emotional cues, pauses and varied delivery.
- Video voiceovers: Produce narration for promotional videos, tutorials, documentaries and explainers.
- Multilingual content: Generate speech for international content workflows across supported languages.
- Creative storytelling: Produce dramatic readings, dialogue scenes and character-driven audio.
- Automated content production: Integrate speech generation into publishing pipelines using CometAPI's asynchronous task workflow.
Limitations and implementation notes
- Gateway-specific character limit: CometAPI announced a 2,500-character limit per request for Eleven v4 on October 10, 2026. This is lower than the native ElevenLabs limit of 10,000 characters. Split longer scripts into coherent segments and verify the current gateway validation rules.
- Expressiveness requires tuning: Highly emotional delivery may not fit every script. Test stability and style settings where supported.
- Pronunciation needs review: Names, abbreviations, numbers and multilingual text may need normalization or revised spelling.
- Segment continuity: When splitting long scripts, use supported
previousTextandnextTextfields where available to help maintain coherent transitions. - Voice availability is gateway-specific: Use the preset voice IDs exposed by CometAPI. Do not assume every voice in the native ElevenLabs library is available through this interface.
- Asynchronous processing: A successful task submission does not mean the audio is ready immediately. Store the task ID, poll for completion and handle failures.
- Not a speech-recognition model: Eleven v4 generates speech from text; it is not intended to transcribe incoming audio.
How to access the Eleven v4 API through CometAPI
The documented endpoint is POST /runwayml/v1/text_to_speech, using Bearer authentication and JSON input. CometAPI returns a task ID that can be used to check status and retrieve the resulting audio.