Eleven v4

Eleven v4 voice clonewhat changed, and how to clone yours

Eleven v4 is ElevenLabs’ newest speech model, launched in September 2026. ElevenLabs says its instant voice clones now capture a voice from as little as 10 seconds of audio, speak 90+ languages with a native accent, and stay the same voice when you regenerate a line. You can clone your voice on it directly at ElevenLabs, or with oVideo Voice Clone, which always speaks with Eleven v4.

Facts from ElevenLabs’ own pagesTwo ways to cloneHonest price table90+ languagesChecked November 20, 2026

Updated

Short answer

Can I clone my own voice with Eleven v4?

Yes. Eleven v4 supports instant voice clones (a short sample, no training) and professional voice clones (a model trained on 30 minutes or more). At ElevenLabs, instant cloning starts on the $6 Starter plan and is the cheapest way to get plain audio. On oVideo, Voice Clone makes an instant clone from 10 seconds to 3 minutes of your voice and reads up to 5,000 characters at a time with Eleven v4, as a 192 kbps MP3. It costs more per minute than ElevenLabs direct, and in return sits in one account with oVideo’s video tools, shows the price on the button and refunds failures automatically.

  • ElevenLabs direct: cheaper per minute, more controls, professional clones
  • oVideo: Eleven v4 by default, 192 kbps on every plan, video tools in the same account
  • Your own voice, or one you have permission to use
  • Accent: native in other languages, yours in your own (per ElevenLabs)

Checked in the oVideo app on November 20, 2026.

What Eleven v4 is

ElevenLabs’ newest model, in its own words

ElevenLabs launched Eleven v4 on September 28, 2026, calling it “our most emotive text-to-speech model yet”, built on “an entirely new architecture”. It comes in two versions: Eleven v4 (eleven_v4) for content such as voiceovers and audiobooks, and Eleven v4 Turbo (eleven_v4_turbo), which ElevenLabs says has a median inference latency of about 100 ms for real time agents.

You direct the delivery inside the text, with tags like [laughs] or [whispers], and ElevenLabs says v4 follows them more reliably than v3, while also saying tags are “not perfect yet”. It improved pronunciation with IPA, keeps long scripts steady across stitched generations, and supports 90+ languages, up from 70+ on Eleven v3.

There are fewer knobs than before: v4 uses Stability and Similarity only. ElevenLabs states that “Style and Speed sliders are not available in Eleven v4, and SSML is not supported”.

For voice cloning

What changed for your clone

  • Shorter samples. The launch post says instant clones can now capture a voice with high fidelity from just 10 seconds of audio. The docs still recommend 1 to 2 minutes for the best result.
  • A closer copy. ElevenLabs says v4 copies timbre, cadence, accent, loudness and even flaws in the recording more faithfully, so “the quality of the source recording matters more than ever”.
  • A native accent abroad. Write in your own language and your accent stays. Write in another language and v4 speaks it fluently with a native accent instead of yours. ElevenLabs is researching a toggle.
  • No drift on retakes. “Redo a line once or fifty times and it’s still the same person speaking”, says the v4 page.
  • Professional clones are back. They were not supported in v3. In v4 they return, with an extra Clone Strength slider.
  • It may sound different from v3. ElevenLabs warns a v4 clone can sound quite different from its v3 version, and that you may still prefer v3 for some voices.

Eleven v3 vs Eleven v4, November 20, 2026

v3 and v4 side by side

From ElevenLabs’ models page, its Eleven v4 page and elevenlabs.io/v4, read on November 20, 2026. Where a page does not compare a point, the cell says so.

PointEleven v3Eleven v4
Languages70+ (models page)90+ (models page; its FAQ names 91)
Instant Voice ClonesSupportedA “significant upgrade”; high fidelity from 10 seconds of audio, per the launch post
Professional Voice ClonesNot supported, per elevenlabs.io/v4Supported again
Accent in another languageThe voice’s own accent carried over; the v4 page calls the new behavior a deliberate changeFluent, native accent in the target language; your accent kept in your own language
Characters per request (API)5,000100,000
Voice settingsNot compared on the pages we readStability and Similarity (plus Clone Strength for Professional clones); no Style or Speed slider; no SSML
Fast variantEleven v3 ConversationalEleven v4 Turbo, about 100 ms median inference

Route 1

Clone your voice on Eleven v4 at ElevenLabs

The steps from ElevenLabs’ Instant Voice Cloning guide, read on November 20, 2026. You need its Starter plan ($6 a month) or higher; professional clones start on Creator.

  1. 01

    Open Voices

    In the ElevenLabs dashboard, select Voices on the left, then click the plus icon.

  2. 02

    Pick Instant Voice Clone

    Choose Instant Voice Clone in the window that opens.

  3. 03

    Upload or record

    Follow the on screen steps. ElevenLabs recommends about 1 to 2 minutes of clean audio; each file needs at least 5 seconds and 10 seconds in total.

  4. 04

    Confirm consent and save

    Name the voice, confirm you have the right and consent to clone it, then click Save voice.

  5. 05

    Use it with Eleven v4

    Under My Voices, click Use voice, pick Eleven v4 as the model and generate.

Route 2

Clone your voice on Eleven v4 with oVideo

oVideo Voice Clone makes an instant clone through ElevenLabs and always speaks with Eleven v4. There is no model to pick and no slider to set. It runs in the browser on a computer or a phone, from Starter at $10 a month.

  1. 01

    Open Voice Clone and tap New voice

    Sign up or log in and open Voice Clone from Tools. Choose Record or Upload.

  2. 02

    Record or upload 30 s to 2 min

    Read the paragraph on screen, or upload a voice memo or a video of you talking (MP4, MOV and more, up to 100 MB). At least 10 seconds of speech is needed.

  3. 03

    Tick the consent box

    Tick “This is my voice, or I have permission to use it” and press Create my voice. It is ready in about a minute.

  4. 04

    Type up to 5,000 characters

    Write in the language you want to hear. The credit price shows on the button: 25 credits per started 30 seconds.

  5. 05

    Download the MP3

    Press Generate voice, listen, and download a 192 kbps MP3. Your last 20 generations stay listed.

Honest comparison, November 20, 2026

ElevenLabs direct vs oVideo Voice Clone

Both use Eleven v4 for your cloned voice. The ElevenLabs column comes from elevenlabs.io/pricing and its help center, read on November 20, 2026; the oVideo column from our app the same day. oVideo is our product, so read its column with that in mind. In short: ElevenLabs direct is cheaper per minute and gives you more controls.

PointElevenLabs directoVideo Voice Clone
EngineEleven v4 (and v4 Turbo, v3 and older models)Eleven v4, always
Clone typeInstant from Starter; Professional from CreatorInstant only
Cheapest plan with cloningStarter $6 a month ($5 billed yearly)Starter $10 a month ($7.50 billed yearly)
About 1 minute of speechAbout $0.20 on Starter, $0.18 on Creator, $0.17 on Pro (plan price divided by the minutes in its table)About $1.25 on Starter, $1.00 on Creator, $0.86 on Pro (50 credits a minute)
Voices you can keep10 on Starter, 30 on Creator, 160 on Pro1 on Starter, 3 on Creator, 5 on Pro
ControlsStability, Similarity, Clone Strength (Professional), Voice Design, a pronunciation dictionary, documented audio tagsNone: text is read as typed with fixed settings
Text per generationUp to 100,000 characters per API request on v4Up to 5,000 characters
Audio quality128 kbps MP3 on Free and Starter; 192 kbps on higher plans192 kbps MP3 on every plan
SampleUpload or record; 1 to 2 minutes recommended; 10 seconds minimum in totalRecord in the browser or upload audio or video up to 100 MB; at least 10 s, 30 s to 2 min best
Price before you clickCredits per character; its FAQ says whether a regeneration is free is shown before generatingThe credit price is on the Generate button
If a generation failsNot stated on the pricing pageCredits are refunded automatically
Model trainingOn by default for non Enterprise accounts; you can turn “Improve the models for everyone” offoVideo has opted out (since November 20, 2026)
Also in the same accountDubbing Studio, Voice Changer, music, sound effects, Studio projects, an APIPut Yourself In, Story to Video, Add AI Voice, subtitles (the clone moves between them as an MP3 today)

List prices in US dollars on November 20, 2026. ElevenLabs’ per minute figures divide each plan’s monthly price by the minutes its own table says the plan includes; its pages do not state a separate rate for Eleven v4, so check in its app. Promotions change: that day its v4 page offered a first month for $1 and its pricing page a Black Friday yearly deal until December 1. oVideo yearly billing brings our minute to about $0.64 to $0.94.

Which one should you pick?

Each route wins its own job

  1. 1The lowest price per minute of plain audio: ElevenLabs direct, from Starter at $6 a month.
  2. 2A professional clone trained on 30 minutes or more of your voice: ElevenLabs Creator or higher; oVideo makes instant clones only.
  3. 3Sliders, a pronunciation dictionary, long requests or an API: ElevenLabs direct.
  4. 4Real time speech for a voice agent: Eleven v4 Turbo through ElevenLabs’ API.
  5. 5Your Eleven v4 voice next to Put Yourself In, Story to Video, Add AI Voice and subtitles, in one account: oVideo Voice Clone, from Starter at $10 a month. The clone moves between tools as an MP3 today.
  6. 6A 192 kbps MP3 on the cheapest plan, the price on the button and refunds on failures: oVideo Voice Clone.

Why it is everywhere

The Eleven v4 launch videos

Eleven v4 is the most watched voice topic on YouTube right now. On November 20, 2026, ElevenLabs’ own launch video, “Introducing Eleven v4 and Eleven v4 Turbo”, had 1,668,188 views after about a month. Creator reviews followed: “Eleven v4 Is INSANE!” (10,912 views), “ElevenLabs v4: How I Clone My Voice for Every Video” and “ElevenLabs v4 Voice Clone: Does Instant Finally Beat Professional?”. Google suggests “eleven v4 pricing”, “eleven v4 free” and “eleven v4 vs v3” as the next questions, which is why this page answers them from ElevenLabs’ own pages.

Better clones

Tips for an Eleven v4 voice clone

Because v4 copies more of the recording, the sample decides most of the result. These tips follow ElevenLabs’ Eleven v4 and Instant Voice Cloning pages. Ready made texts are on voice cloning script to read.

  • Record in a quiet, soft room with no music, TV or second voice. v4 will copy hiss and echo too.
  • Keep one speaking style the whole time. ElevenLabs recommends a single style for v4 training audio.
  • Aim for 1 to 2 minutes. ElevenLabs says more than 3 minutes adds little and can even hurt.
  • Keep the level steady: not too quiet, not clipping.
  • Record in the language you will use most; that is where your accent is kept.
  • Write the text the way you speak it. ElevenLabs says ellipses add pauses and capitals add emphasis on v4.
  • On oVideo, a failed or too short sample is refunded, and one re-record within 24 hours is free.

oVideo Voice Clone

The facts, on Eleven v4

Model
Eleven v4 for every generation; instant clone
Sample
At least 10 s of speech, best 30 s to 2 min; first 3 min used
Record or upload
In the browser, or MP3, M4A, WAV, MP4, MOV and more, up to 100 MB
Ready
In about a minute
Text
Up to 5,000 characters per generation, in any of the 90+ languages Eleven v4 supports
Output
MP3, 44.1 kHz, 192 kbps
Price
New voice 50 credits once; speech 25 credits per started 30 s; failures refunded
Voices per plan
Starter 1, Creator 3, Pro 5
Training
oVideo has opted out of ElevenLabs model training (since November 20, 2026)
Not included
Professional clones, sliders, singing, real time, use inside other oVideo tools

Checked in the oVideo app on November 20, 2026.

Privacy and consent

Your voice, your permission

Both routes ask for consent: ElevenLabs asks you to confirm the right and consent to clone the voice, and oVideo asks you to tick “This is my voice, or I have permission to use it”. ElevenLabs blocks the cloning of celebrity and other high risk voices.

On ElevenLabs direct, its help center says customer data is used to improve its models unless you switch off “Improve the models for everyone” under Data use. oVideo has opted out of ElevenLabs model training (since November 20, 2026), so voices you clone are not used to train ElevenLabs models.

On oVideo your sample is stored privately and deleted together with the voice when you delete it; the MP3s you made are separate files, so keep your downloads. Commercial use of the audio is allowed under our Terms, provided you had the right to use the voice you cloned.

Checklist

What to check before you clone on Eleven v4, answered for oVideo

Which model reads my text?

Eleven v4, every time.

Is it free?

No. Creating a voice needs Starter ($10 a month) or higher.

Is it cheaper than ElevenLabs?

No. ElevenLabs direct costs less per minute of audio.

Professional clone?

No, instant clones only.

Sliders or tags?

No sliders. Text is read as typed; tags are not a tested feature.

Audio quality?

MP3, 44.1 kHz, 192 kbps, on every plan.

What if it fails?

The credits come back automatically.

Used for training?

oVideo has opted out of ElevenLabs model training (since November 20, 2026).

Limits and prices checked in the app on November 20, 2026.

Answers

Frequently asked questions

What is Eleven v4?
Eleven v4 is ElevenLabs’ text to speech model launched on September 28, 2026, built on a new architecture. ElevenLabs calls it its most emotive model, says it supports 90+ languages, and offers a faster variant, Eleven v4 Turbo, for real time voice agents. It also brought back Professional Voice Clones, which v3 did not support.
What is Eleven 4?
People often mean Eleven v4, the ElevenLabs speech model released in September 2026. The model ID is eleven_v4; the faster version is eleven_v4_turbo.
Can I clone my voice with Eleven v4?
Yes. ElevenLabs says Eleven v4 works with both Instant Voice Cloning (a short sample, no training step) and Professional Voice Cloning (a model trained on longer recordings). You can do it on ElevenLabs from its Starter plan, or on oVideo Voice Clone, which makes an instant clone and always speaks with Eleven v4.
How do I clone my voice on Eleven v4?
At ElevenLabs: open Voices, click the plus icon, pick Instant Voice Clone, upload or record about 1 to 2 minutes, confirm you have the right and consent to clone the voice, save it, then pick Eleven v4 when you generate. On oVideo: open Voice Clone, record 30 seconds to 2 minutes or upload a clip, tick the consent box, wait about a minute, then type your text; it is always read with Eleven v4.
How much audio does an Eleven v4 voice clone need?
ElevenLabs’ launch post says Instant Voice Clones can now capture a voice with high fidelity from just 10 seconds of audio, while its docs still recommend 1 to 2 minutes and say more than 3 minutes adds little. Professional clones need 30 minutes or more. oVideo needs at least 10 seconds of speech, works best with 30 seconds to 2 minutes and uses the first 3 minutes.
Is Eleven v4 free?
ElevenLabs’ free plan (10,000 credits a month on its pricing page, read November 20, 2026) includes text to speech, and its launch post invites you to create a free account to try Eleven v4. Voice cloning is not on the free plan: its pricing page lists Instant Voice Cloning from Starter at $6 a month, and the free plan has no commercial license. Promotions change often, so check its pricing page. oVideo has no free voice cloning either; it starts at $10 a month.
How much does ElevenLabs cost a month?
On its pricing page on November 20, 2026: Free $0, Starter $6, Creator $22 (a Black Friday yearly deal of $8.25 a month for the first year runs until December 1), Pro $99, Scale $299 and Business $990, with Enterprise on request. Instant Voice Cloning starts at Starter and Professional Voice Cloning at Creator.
Is it cheaper to use Eleven v4 through oVideo than directly at ElevenLabs?
No. For audio alone, ElevenLabs direct costs less per minute: about $0.17 to $0.20 by its own pricing table, against about $0.64 to $1.25 on oVideo depending on plan and billing. oVideo makes sense when you also want its video tools in the same account, a price shown on the button and automatic refunds for failures.
Is Eleven v4 better than v3?
ElevenLabs says v4 “is a net upgrade over Eleven v3, delivering better results in almost every case” and recommends testing it with your own voices. It also warns that a v4 clone may sound quite different from its v3 version, because v4 copies the source voice more faithfully, and that you may still prefer how a voice sounded in v3.
Will my Eleven v4 clone keep my accent?
Per ElevenLabs: in the same language as your sample, yes, your accent is kept. In a different language, v4 speaks fluently with a native accent instead of carrying yours over. ElevenLabs says it is exploring a toggle for this and that audio tags may help guide an accent, with results that vary. We have not measured it ourselves.
Which languages does Eleven v4 speak?
ElevenLabs says 90+ and names 91 on its Eleven v4 page, from Afrikaans and Arabic to Turkish, Vietnamese, Welsh and Zulu. Its pricing table still shows 74 languages per plan, so the model page is the newer source.
What is the difference between instant and professional voice cloning?
An instant clone imitates a short sample on the fly, with no training step: ready in seconds at ElevenLabs and in about a minute on oVideo. A professional clone trains a dedicated model on 30 minutes or more of your voice; ElevenLabs says fine tuning usually takes 3 to 6 hours. oVideo makes instant clones only.
Does Eleven v4 have speed, style or emotion controls?
ElevenLabs says Style and Speed sliders are not available in Eleven v4 and SSML is not supported. Delivery is directed with the text itself and with audio tags such as [whispering] or [laughing], which ElevenLabs says are not perfect yet. On oVideo there are no sliders at all, and tags are not a feature we have tested.
Which ElevenLabs model is best?
For quality, ElevenLabs points to Eleven v4 for content, audiobooks and voiceovers, and to Eleven v4 Turbo for real time agents. Its own page still says a few edge cases may suit another model, so test a short line first.
Can I use an Eleven v4 voice without an ElevenLabs subscription?
Yes, through apps that run on ElevenLabs, such as oVideo Voice Clone. You still pay that app instead: oVideo needs its Starter plan ($10 a month) or higher to create a voice, and it costs more per minute than ElevenLabs direct.
Can I use my Eleven v4 voice clone commercially?
At ElevenLabs, its help center says every paid plan includes a commercial license (Beta Services aside) and the free plan has none. On oVideo, our Terms allow commercial use of audio made with Voice Clone, provided you had the right to use the voice you cloned. This is not legal advice.
Is ElevenLabs safe to use? Is my voice used for training?
ElevenLabs’ help center says it uses some customer data to improve its models unless you turn off “Improve the models for everyone” in Data use; after that, new data is not used for training. oVideo has opted out of ElevenLabs model training (since November 20, 2026), so voices you clone are not used to train ElevenLabs models.
Is voice cloning illegal?
Cloning your own voice, or a voice you have permission to use, is what these tools are for. ElevenLabs asks you to confirm the right and consent to clone, its use policy forbids cloning a living person without explicit consent, and oVideo’s Terms say the same. Laws differ by country, so this is not legal advice.
Can Eleven v4 sing?
ElevenLabs lists a singing tag only as experimental. oVideo Voice Clone speaks and does not sing.
Can I clone my voice in real time with Eleven v4?
Eleven v4 Turbo is ElevenLabs’ real time variant for voice agents, through its API. oVideo is not real time: you type, press Generate voice and get an MP3 a moment later.
Why does my Eleven v4 clone sound different from my v3 clone?
ElevenLabs says v4 reproduces the source voice much more faithfully, including its accent, loudness and flaws in the recording, so it can sound quite different from v3. Its advice: use clean audio in a single speaking style, and compare both versions with your own text.
What is the cheapest way to clone my voice with Eleven v4?
For plain audio, ElevenLabs’ own Starter plan at $6 a month with about 30 minutes of speech, by its pricing table. If you want a 192 kbps MP3 on the cheapest plan, video tools in the same account and automatic refunds, oVideo starts at $10 a month and costs more per minute.

Your voice on Eleven v4, in about a minute

Record or upload 30 seconds to 2 minutes, tick the consent box, then type anything in 90+ languages and download a 192 kbps MP3. From Starter at $10 a month.

Voice cloning guides

Related