by ElevenlabsReleased 7 October 2026
Reads English text aloud at ultra-low latency for live conversation, taking 30,000 characters a request, where Turbo v2 is the earlier low-latency generation and Flash v2.5 adds 32 languages
You pay for what each request uses. No subscription, no minimum.
One paragraph read aloud, the ordinary use
Every rate
| Meter | ↑ you send | ↓ comes back |
|---|---|---|
| Characters | 13 313,42 DA |
per 1M characters
One company runs this model. OpenDunes sends every request to it. DA per 1M characters: ↑ what you send.
| Provider | ↑ Sent | Latency | Uptime |
|---|---|---|---|
| ElevenLabs | 13 313,42 | … | … |
You pay the price of the provider that served your request. Prices read 15:19 · speed over the last 30 minutes · uptime over 3 days.
Swap the base URL and the key. Your OpenAI code runs unchanged.
import os
from openai import OpenAI
client = OpenAI(
base_url="https://opendunes.com/v1",
api_key=os.environ["OPENDUNES_API_KEY"],
)
with client.audio.speech.with_streaming_response.create(
model="elevenlabs/eleven-flash-v2",
input="مرحبا بكم في متجرنا. التوصيل مجاني في وهران.",
voice="george",
) as resp:
resp.stream_to_file("speech.mp3")Runs this request once on your balance.
About 7,99 DA per paragraph read aloud. ↑ 13 313,42 DA per 1M characters.
No. You pay per use from your balance, with no subscription. Try it in Chat at the same price.
Reads English text aloud at ultra-low latency for live conversation, taking 30,000 characters a request, where Turbo v2 is the earlier low-latency generation and Flash v2.5 adds 32 languages.
Eleven Flash v2.5 costs the same for the same paragraph read aloud.
Deposit dinars with your Edahabia or CIB card. Every model, Eleven Flash v2 included, is paid from that balance. No foreign card, no conversion.
Elevenlabs names no languages for Eleven Flash v2. Arabic, French and Darja: not stated.
21 voices. Cloning from a sample: not stated.
Elevenlabs. Released 7 October 2026.
Eleven v4 Turbo turns text into expressive speech with low latency, keeping v4's audio tags and delivery while generating faster across more than 90 languages
Eleven v4 is ElevenLabs' most expressive voice model: you write cues into the text to steer emotion and delivery, and get speech back in 90+ languages. Requests take up to 10,000 characters, twice what v3 accepts. Pick it for character voiceovers, emotional dialogue and audiobooks.
Transcribes audio in 90+ languages with word-level timestamps, optional speaker diarization and audio-event tagging, for any subject where Scribe v2 Medical is the one tuned for clinical terms
Eleven Flash v2.5 reads text aloud in 32 languages and starts speaking in about 75 milliseconds, which suits voice assistants, chatbots and big batches of audio. One request can hold up to 40,000 characters. For finished narration or acted emotion, Multilingual v2 and Eleven v4 are the models built for it, not for live speed.
From what its maker and two benchmark sites publish, and where it stands on OpenDunes. Read 7 Nov 2026.
Clones a voice No.
21 voices
Answer a live English caller with speech that starts at once
Speak a 30,000-character English document in one request
Choose from 21 voices for an English-only live assistant