ElevenLabs is best known for AI voices. You type or paste a script, pick or design a voice, and it produces speech in many languages. Over time the company has added voice cloning, sound effects, music, dubbing, speech to text, an API and a platform for voice agents.
This review summarises what ElevenLabs’ own pricing page, help centre and documentation say about its products, credit system, voice cloning rules and data use. It is for people deciding whether to pay for it, such as podcasters, video teams, course creators and developers.
How we researched this: we did not run hands-on tests for this article. It is based on the vendor’s own documentation and public sources, checked on 11 November 2026. Features and prices change, so confirm details on the vendor’s site before you buy.
Key takeaways
- ElevenLabs plans run from a free tier (10,000 credits a month) up to Business and custom Enterprise, with the commercial licence starting on the Starter plan.
- Credits (previously called characters) are the unit of usage. Self-serve plans let up to two months of unused credits carry over, which is less common than strict monthly resets.
- Professional Voice Cloning needs the Creator plan or above, and you can only clone your own voice, verified through a process, even if someone else consents.
- By default, ElevenLabs uses certain data you provide to improve its models, and you can switch this off in your account settings. Enterprise customers are excluded by default.
- Free plan output can be used only non-commercially and needs attribution, according to the help pages we read.
What ElevenLabs is
ElevenLabs’ pricing page lists a set of creative tools on its self-serve plans: text to speech, speech to text, sound effects, Voice Design, music, Studio projects, Productions, dubbing and image and video generation. The company also sells API access and a separate platform for conversational agents, each with its own pricing page. This review focuses on the creative plans.
The voice model
ElevenLabs’ page on Eleven v4 describes it as its “most emotive voice model”, built on a new architecture. It says v4 supports inline direction through audio tags such as [laughs] and [whispers], keeps pacing consistent across long scripts, supports multiple speakers in one script and covers speech in more than 90 languages. It says Professional Voice Clones are available again in v4, after not being supported in v3. A faster variant, Eleven v4 Turbo, is listed for real-time use through the API and the agents platform, with a stated median inference latency of about 100 milliseconds. The page gives no release date, and these are the vendor’s claims, which we have not verified.
Who it is for
Based on the documented features, ElevenLabs suits people who need spoken audio at scale or with control over tone: video creators adding narration, educators producing course audio, publishers experimenting with audiobooks, product teams adding voice to apps through the API, and businesses building voice agents. Our beginner’s guide to AI voice and text to speech explains the underlying concepts if you are new to the area.
It is less obviously suited to people who want to clone another person’s voice, because the rules described below are strict, or to teams that cannot accept data being used for model improvement without changing a setting.
Key features (documented)
| Feature | What the vendor’s pages say |
|---|---|
| Text to speech | Eleven v4 with audio tags, multi-speaker scripts and 90+ languages. |
| Instant Voice Cloning | Creates a clone from short samples without training a custom model. About 1 to 2 minutes of good audio is recommended. |
| Professional Voice Cloning | Trains a dedicated model on 30 to 180 minutes of audio, typically taking 3 to 6 hours, with a verification process. |
| Voice Design | Creates voices from text descriptions. |
| Pronunciation Dictionary | Supports IPA, according to the v4 page. |
| Sound effects and music | Listed in the core tools. Music commercial use is listed from Starter. |
| Dubbing Studio, Studio projects | Listed on paid plans, with Studio project limits by plan. |
| API and agents | Separate pricing pages exist for the API and for ElevenAgents. |
Voice cloning in detail
ElevenLabs’ voice cloning documentation separates two types. Instant Voice Cloning is quick and may struggle with very unique voices or uncommon accents. Professional Voice Cloning is higher effort and aims for higher realism. The page recommends MP3 at 192kbps or higher, clean single-speaker audio and consistent tone. Professional cloning needs the Creator plan or above, with slots by plan: none on Free and Starter, one on Creator and Pro, three on Scale and ten on Business, according to the documentation.
Pricing model
Checked on 11 November 2026. All prices below are monthly billing in US dollars as shown on the creative plans page, and the page says prices exclude taxes. A yearly billing toggle exists, but we could not read the yearly prices, so confirm them on ElevenLabs’ site. The page also lists included minutes, but the figures were hard to match to columns in what we could read, so we omit them.
| Plan | Price shown | Credits per month | Notable additions listed |
|---|---|---|---|
| Free | $0 | 10,000 | 3 Studio projects and the core tools |
| Starter | $1 first month, then $6 | 30,000 | Commercial licence, Instant Voice Cloning, 20 Studio projects, Dubbing Studio |
| Creator | $11 first month, then $22 | 121,000 | Professional Voice Cloning and additional credits |
| Pro | $99 | 600,000 | 44.1kHz PCM audio via API and 192kbps audio |
| Scale | $299 | 1.8 million | 3 workspace seats, team collaboration, 3 Professional Voice Clones |
| Business | $990 | 6 million | Low-latency text to speech, 10 Professional Voice Clones, 10 seats |
| Enterprise | Custom | Custom | Custom terms, DPA and SLA assurances, SSO, priority support |
How credits behave
ElevenLabs’ billing page says credits were previously called characters, with unchanged value. The number of credits a generation uses depends on your plan, whether you use the website or the API, the product and the model. On self-serve monthly or annual plans, your quota resets monthly and up to two months’ worth of unused credits carry over. If you downgrade or cancel, the change takes effect at the end of your billing cycle and unused credits do not roll over.
- New subscriptions can prepay for extra credits (pay as you go) that last 12 months and are used once your monthly quota runs out. This is not available on Enterprise or legacy subscriptions.
- Refunds are available if you request one within 14 days of payment and no credit quota was used in that period.
- Our guide to reading an AI tool’s pricing page explains how to translate credits into the cost of your real workload.
Strengths
- Wide product range. Speech, cloning, effects, music, dubbing and an API sit under one account.
- Low entry points. A free tier and a low-priced Starter plan make trial cheap, though Starter is a promotional first month.
- Credit carry-over. Up to two months of unused credits roll over on self-serve plans.
- Clear cloning boundaries. Professional cloning requires verification and limits you to your own voice.
- Documented opt-out. The data-use setting is described step by step in the help centre.
- Enterprise options. Higher tiers list DPA, SLA, SSO and HIPAA business associate agreements.
Limits and cautions
- Credit use is hard to predict. Because cost depends on product, model and plan, you need to estimate with your own scripts.
- Free plan restrictions. The help centre says free plan content can be used non-commercially and needs attribution.
- Cloning limits. You can only clone your own voice for Professional Voice Cloning, even with someone else’s consent. For another person’s voice, ElevenLabs describes a sharing route where the owner creates and verifies the clone on their own account and shares it.
- Data use is on by default. You must switch it off yourself on non-enterprise accounts.
- Pricing pages are moving. The pricing page now uses the name ElevenCreative, and older documentation refers to legacy plans, so check which applies to you.
- Reading limits. Several yearly prices and the included minutes could not be read reliably in our research.
Privacy and data
ElevenLabs’ help centre says it uses certain data you provide to improve its audio models for everyone, and does not use your data to clone your voice without consent or to create advertising profiles. Enterprise customers are excluded by default, except where their data is needed to provide services. To opt out, go to your profile menu, then Terms and privacy, then Data use, and turn off the “Improve the models for everyone” toggle. Data you submit afterwards will not be used to train models. The page did not mention zero retention, so we could not confirm it; the privacy policy is the referenced source.
If you plan to upload customer recordings, employee voices or confidential scripts, read our guide to protecting company data when using AI tools and decide on the opt-out before you upload anything.
Responsible use of synthetic voices
Synthetic audio can mislead listeners if it is not labelled. Our article on spotting AI-generated content covers the audience side, and your own policy should cover disclosure and consent for any voice you use. This is general information, not legal advice.
Alternatives
We did not research competing voice tools in depth, so we do not rank them. If you need a presenter on screen rather than only a voice, our HeyGen review covers an avatar video tool. If you are producing video, our Runway review covers a video platform that, according to Runway’s changelog, added ElevenLabs V4 for audio generation on 29 September 2026. Use our guide to comparing AI tools side by side fairly to build a fair test script.
Who should and should not use it
| Best for | Look elsewhere if |
|---|---|
| Creators and teams who need narration, dubbing or effects in one place | You need to clone voices that are not your own |
| Users who want to start cheaply and scale credits up | You cannot accept default data use and do not want to manage the opt-out |
| Developers building voice features through the API | You need fixed, easily predictable per-minute costs |
| Businesses needing enterprise terms such as DPA and SLA | You want a free plan for commercial work, which it does not allow |
Questions to ask before buying
- How many credits will our typical script use in the model and product we plan to use?
- Do we need Professional Voice Cloning, and does our voice talent satisfy the verification rules?
- Will we switch off the “Improve the models for everyone” setting, and is that enough for our compliance needs?
- Do we need commercial rights from day one (Starter or above)?
- Do we need enterprise terms, and who handles the contract?
Before paying, follow our guide to testing an AI tool before you pay for it, with the same script run on each tool you shortlist.
Frequently asked questions
Is ElevenLabs free?
There is a free plan with 10,000 credits a month and core tools. Free plan output is limited to non-commercial use with attribution, according to the help centre.
Can I clone any voice?
No. Professional Voice Cloning only allows your own voice and requires verification, even if someone else gives permission. Instant Voice Cloning has its own requirements, so read the documentation.
Do unused credits roll over?
On self-serve plans, up to two months’ worth of unused credits carry over. If you downgrade or cancel, unused credits do not roll over.
Does ElevenLabs train on my data?
By default it uses certain data you provide to improve its audio models, and you can opt out in your account settings. Enterprise customers are excluded by default.
Can I get a refund?
The billing page says you are eligible if you request within 14 days of payment and no credit quota was used in the period.
Are AI voices always accurate?
Not necessarily. Mispronunciations and wrong emphasis can happen, so listen to every output. See our guide to checking AI output for mistakes.
Conclusion
From its own documentation, ElevenLabs is a broad audio platform with a clear credit model, carry-over on self-serve plans, strict rules on professional voice cloning and a default data-use setting you can turn off. The things to check are credit use for your workload, the commercial licence tier and your comfort with the data defaults.
Start with our guide to choosing an AI tool, then confirm every price on ElevenLabs’ own pricing page.
Sources
- ElevenLabs pricing page, 11 November 2026
- ElevenLabs billing documentation, 11 November 2026
- Eleven v4 and Eleven v4 Turbo, 11 November 2026
- Voice Cloning overview (ElevenLabs documentation), 11 November 2026
- Can I create a Professional Voice Clone of someone else’s voice? (ElevenLabs Help Centre), 11 November 2026
- Is my data used to improve ElevenLabs’ AI models? (ElevenLabs Help Centre), 11 November 2026
- Runway product updates and changelog, 11 November 2026
