ElevenLabs
Eleven v4
elevenlabs/eleven-v4
ElevenLabs' most expressive text-to-speech model, with inline audio tags for emotion and delivery, support for 90+ languages, and long inputs per request.
RoleThe expressive voice
$80
In / M
Free
Out / M
—
Context
- AA intelligence index
- Not on listing
- Inputs
- text
- Outputs
- speech
- Tools
- Not listed
- Structured output
- Not listed
- Reasoning
- Not listed
- Weights
- Not verified
- Listed
- Oct 7, 2026
01 · What it's good at
Use it for this.
Three jobs this model is built for, and the ones to give to something else.
01
Voiceovers and narration
02
Spoken replies with tone control via audio tags such as [warmly]
03
Multilingual voice output
Skip it for
- Transcribing audio. Use a transcription model
- The lowest-latency replies. Eleven v4 Turbo is the faster sibling
02 · Example tasks
What you'd actually send it.
Real inputs, not a benchmark. Paste one into the snippet below to try it.
You send
[cheerfully] This week we shipped CSV export and fixed Safari sign-up.
Why this model
Audio tags set the delivery.
03 · Your stack
Where it sits: Draw, speak, listen.
The last step of a voice pipeline: transcription in, chat model in the middle, speech out.
Layer 01
Retrieve
Layer 02
Plan & reason
Layer 03
Do the volume
Layer 04
Gate & route
Layer 05 · this model
Draw, speak, listen
04 · Wire it up
One request. Paste and run.
Uses OpenRouter, so one key works for every model on this site. Set OPENROUTER_API_KEY and swap the prompt for one of the tasks above.
import { writeFile } from "node:fs/promises";
const res = await fetch("https://openrouter.ai/api/v1/audio/speech", {
method: "POST",
headers: {
Authorization: `Bearer ${process.env.OPENROUTER_API_KEY}`,
"Content-Type": "application/json",
},
body: JSON.stringify({
"model": "elevenlabs/eleven-v4",
"input": "[warmly] Welcome back. Your report is ready.",
"voice": "george",
"response_format": "mp3"
}),
});
// The response body is the audio file itself.
await writeFile("reply.mp3", Buffer.from(await res.arrayBuffer()));
05 · Watch out for
Gotchas.
Ask the chat model for plain spoken sentences: no markdown, lists, or emoji.
Voices are listed per model in the catalog (for example george, sarah, adam).
Sources
Lessons are written from each model's listing and the linked docs. We don't publish our own benchmark numbers; the index above is Artificial Analysis, as shown on OpenRouter.
Keep learning