Listing checkedFreemiumPaidContact salesElevenLabs
Text to speech, voice cloning, dubbing and voice agents on one credit balance, in a browser studio and behind a REST API.
- Ratings
- No member reviews yet
- Price
- From $6/mo
- Last updated
Key takeaways
- ElevenLabs is not a single text-to-speech box. The documentation splits it into ElevenCreative (a no-code browser studio), ElevenAgents (voice and chat agents) and ElevenAPI (REST plus Python and TypeScript SDKs). Speech generation, transcription, voice cloning, dubbing, music and sound effects all draw on one credit balance.
- The product has tilted toward business deployments. The company says it prioritised enterprise adoption of ElevenAgents and international expansion through 2026, and its February 2026 funding announcement named Deutsche Telekom, Square, the Ukrainian Government and Revolut as customers. Those are company statements, not verified deployments. Solo creators still get the same studio.
- Pricing you find elsewhere is likely stale. As of August 2026 the live plans run Free at $0, Starter at $6, Creator at $22 (shown at $11 for the first month only), Pro at $99, Scale at $299 and Business at $990 a month, with Enterprise on custom terms. Older listings quoting $5, $11 recurring, $330 and $1,320 are out of date.
- Third-party estimates put elevenlabs.io somewhere between 33 and 35 million monthly visits in June 2026, with two independent estimates disagreeing on the exact figure and the month. Direct traffic accounts for roughly 43%, the pattern of a tool people return to, not one they discover.
- The biggest caveat is billing. Trustpilot sat at 3.2 out of 5 from 989 reviews in June 2026 while G2 sat at 4.5 from 1,130, and the gap is almost entirely subscriptions, refunds and credits. Cancelling forfeits unused credits at the end of the billing cycle, and a refund requires a request within 14 days with none of that period's quota used.
What is ElevenLabs?
ElevenLabs generates and edits audio in a browser and behind an API. You paste text and get speech, upload a recording and get a transcript, hand it a video and get a dubbed version, or describe a sound and get an effect. The homepage positioning as of August 2026 reads "Free AI Voice Generator & Voice Agents Platform". That covers both ends of the product, a studio for people making media and an agent platform for people building software that talks.
The distinguishing bit is that it runs primarily on ElevenLabs' own audio models and one shared credit balance. Text to Speech, Speech to Text, Dubbing, Voice Design, Music, Sound Effects and Studio productions sit in ElevenCreative. ElevenAgents adds a visual builder for voice and chat agents with programmatic controls underneath. ElevenAPI exposes the same capabilities through REST with official Python and TypeScript SDKs, so you can move a workflow prototyped in the browser into a pipeline without changing vendors. The February 2026 announcement named an Eleven v3 Conversational model upgrade with improved turn-taking for agents.
Read the language claims carefully, because two different numbers circulate. Current marketing advertises 10,000+ voices and 70+ languages across the platform. The text-to-speech documentation separately specifies 32 languages. They are not interchangeable. The higher figure covers the platform broadly, the lower one covers TTS specifically. Behind the product is a private, independent company that raised a $500 million Series D in February 2026 at an $11 billion valuation, led by Sequoia Capital.
How it works
Create an account
Open the web application and sign up. New accounts land on the Free tier automatically, with no credit card required. That tier is permanent, not a countdown trial, though its output carries non-commercial and attribution conditions.
Choose a workflow
In ElevenCreative, pick Text to Speech, Dubbing, Voice Design, Sound Effects, Music, Studio productions or the transcription tools. In ElevenAgents, open the visual agent builder instead. That choice decides how fast your credits drain, because each capability charges on a different unit.
Supply content and direct the result
Paste your script, or upload the source audio or video. Select a voice from the library or create one you are authorised to use, then adjust delivery, language and the other controls. Dubbing works from source audio and attempts to hold the original speaker's voice, emotion and timing across the translation.
Export or integrate
Preview and edit in the browser, then download or publish the project output. Or skip the studio entirely. Call the REST API through the Python or TypeScript SDK and embed generation inside your own application. API calls draw on the same credit balance as the browser work.
Features
Text to Speech
The core capability, and the one reviews consistently name as the strength. You get intonation, pacing and emotional variation instead of flat readout, multiple voice styles through the API, and documented support for 32 languages. Billing is one credit per input character, so long-form narration is the fastest way to burn a monthly allowance.
Voice Cloning and Voice Design
Two different jobs. Cloning builds a synthetic voice from a sample you have authorisation to use. Availability and commercial rights vary by plan. Voice Design creates a voice from scratch instead of selecting one from the library, and it appears in the current Free plan feature list. The terms restrict unauthorised cloning, so the sample you supply is your responsibility.
Dubbing
Localises video and film into other languages and tries to keep the original speaker's voice, emotion and timing, across the 70+ languages the platform markets. Charging is by source-audio minute, not by characters, so a long video costs the same whether it's dense with dialogue or mostly silence.
Speech to Text, Voice Isolation and Voice Changer
The audio-in side of the platform. Speech to Text transcribes recordings and bills by processed audio, not by characters. Voice Isolation separates a voice from background material, which is the cleanup step podcast and media workflows usually need first. Voice Changer performs voice-to-voice transformation. The retrieved product material does not specify technical limits for isolation, so try it on your worst recording before you trust it.
ElevenAgents
A visual builder plus programmatic controls for voice and chat agents, aimed at customer support, sales and workflow automation. This is the part of the product the company has pushed hardest at enterprise buyers, and the February 2026 Eleven v3 Conversational upgrade targeted turn-taking, the thing that makes phone agents feel like an interruption or a conversation.
Text to Sound Effects and Music
Generates non-speech audio from a prompt, billed per generation, not by length. Both appear in the current Free plan feature list, so they're cheap to evaluate before you commit to a tier.
ElevenAPI and the SDKs
REST access with official TypeScript and Python SDKs, documentation and examples across TTS, transcription, dubbing, music, effects and agents. Rate limits, quotas and retention behaviour vary by plan and configuration. Don't assume the studio experience translates; check the live documentation against your own throughput.
What people use it for
- Narrating multi-character audiobooks and personalised long-form audio.
- Producing voiceovers for video, advertising and media projects.
- Dubbing films and video into other languages while holding the original delivery.
- Transcribing and cleaning up podcast or spoken-word recordings.
- Building AI phone agents for support, sales and inbound workflows.
- Adding speech and conversational interaction to AI assistants and education products.
- Embedding audio generation inside publishing, media-production and creator platforms through the API.
It's a poor fit for confidential media or sensitive customer calls unless you configure the enterprise privacy controls first, because non-Enterprise data may be used to improve the models by default. And it is not a route to cloning a voice you do not have permission to use.
Who it’s for
Pricing
Official pricingIndividual plans
Free
$0
per month
10,000 credits a month, no card required; output is non-commercial and requires attribution
Starter
$6
per month
30,000 credits a month; the cheapest tier with commercial rights
Creator
$22
per month
121,000 credits a month; $11 covers the first month only, then it reverts to $22
Pro
$99
per month
600,000 credits a month for sustained long-form and API work
Scale
$299
per month
1,800,000 credits a month and three seats
Business plans
Business
$990
per month
6,000,000 credits a month and ten seats
Enterprise
Custom
contact sales
Custom credits and seats, Zero Retention Mode, and data excluded from model training by default
Everything runs on one credit balance shared across ElevenCreative, ElevenAgents and ElevenAPI, and each capability meters differently: text to speech charges one credit per input character, speech to text charges by processed audio, music and sound effects charge per generation, and dubbing charges by source-audio minute. There is no single cost for "one generation" because it depends on the product, the model and the output length. Monthly credits reset, and unused credits roll over for up to two months. The Free tier is permanent, not a trial, with 10,000 credits a month and no card required, but its output is non-commercial and requires attribution, so commercial publishing means a paid plan. Watch three written terms before you subscribe. The Creator plan's $11 price is a 50% first-month promotion and reverts to $22. Cancellation takes effect at the end of the current billing period, after which the account drops to Free and unused credits are lost, though the cancellation can be reversed before the cycle ends. Refunds require a request within 14 days of payment with none of that period's credit quota used, sent from the account email to team@elevenlabs.io; otherwise fees already paid are not refunded, with EU consumer withdrawal rights preserved separately. Website subscriptions bill through Stripe, and app-store purchases follow different rules. Annual-billing discounts, top-up pack pricing, per-tier custom-voice slots, audio-quality differences, SSO and admin controls and Enterprise SLA terms were not exposed on the pricing page when we checked in August 2026. Prices are the live figures as of August 31, 2026.
Traffic & reach
Third-party estimates put elevenlabs.io in the range of 33 to 35 million visits for June 2026, and the two estimates we have do not agree on the exact number or cleanly on the month, so trust the range, not either figure. The shape tells you more than the size. Roughly 43% of visits arrive direct and 28% through organic search, with about 4.6 million brand searches a month. That is a bookmarked working tool, not a page people stumble onto. The United States and India are the two largest sources at about 17% and 14%. None of this is an audited company metric.
Visits, June 2026
33.6M
▼ 4.3% month over month
Nov '25Jun '26
Top countries
- United States17%
- India14%
- Philippines4.1%
- Brazil4.1%
- Pakistan3.8%
Traffic sources
- Direct43%
- Organic search28%
- Paid search9.5%
- Organic social5.9%
- AI assistants3.0%
- Referrals2.9%
- Display ads2.5%
- Affiliates2.4%
- Paid social1.7%
- Email1.1%
- Visits, last 3 months
- 102.0M
- Avg. visit
- 4m 50s
- Pages per visit
- 6.2
- Bounce rate
- 41.3%
- Brand searches / mo
- 4.6M
User reviews
Write a reviewNo reviews yet
Used ElevenLabs on something real? Write down what you made, what it cost and what the trade-off was. The next person can decide in about a minute from that.
Be the first to reviewWhat to know before you commit
Voice quality is the consistent strength. Trustpilot's March 2026 summary reports users praising natural-sounding voices, the breadth of the library and often helpful staff. Independent review summaries single out emotionally expressive TTS, cloning from short samples and low-latency streaming. G2 aggregation reported 4.5 out of 5 from 1,130 reviews in June 2026, and other analyses put it around 4.5 to 4.6, with counts varying by date and source. Professional users rate the output highly, and consistently.
The billing complaints are a pattern, not noise. Trustpilot sat at 3.2 out of 5 from 989 reviews in June 2026, well below the G2 figure, and the summaries point at the same handful of issues: confusing subscription mechanics, unexpected charges, credits lost on cancellation, fast credit burn on long-form projects, and disputes over refunds and account access. The published policy is clearer than the complaints suggest, which usually means people meet the terms after purchase instead of before. Read the refund window and the cancellation forfeiture before you pay.
Usability and support draw mixed reports. Reviewers describe a clunky interface when they need precise control over pauses or tone, and difficulty reaching a human when something goes wrong. Others report responsive staff. These are aggregate summaries, not a representative survey, and current Capterra, App Store and Google Play counts were not established when we checked.
Voice cloning carries a consent problem the product cannot fully solve. A May 2026 secondary investigation says ElevenLabs voices have appeared in non-consensual audio deepfakes, and characterises consent mechanisms as present but hard to enforce. The company's terms prohibit misuse and its product materials emphasise voice consent. No primary case-by-case incident register turned up in the sources we checked. If you are cloning a voice for commercial work, the publicity-rights and consent exposure is yours to clear.
Data use is a default, and a setting. For non-Enterprise accounts, ElevenLabs says certain submitted data may be used to improve its audio models by default, with an opt-out under Terms and privacy > Data use ("Improve the models for everyone"). Enterprise data is excluded from training by default and Zero Retention Mode is available. The terms also say the company will not commercialise a user's voice on a standalone basis without permission. For confidential material, get retention, training exclusion and regional processing into the contract, not into a toggle.
The growth figures are the company's own. ElevenLabs says it closed 2025 above $330 million in ARR and that the February 2026 Series D valued it at $11 billion, and it names Deutsche Telekom, Square, the Ukrainian Government and Revolut as customers. None of the revenue or customer claims are independently audited. Separately, third-party outage trackers reported 154 or more incidents over nine months from May 2025 in one count and 74 in another, but those are monitoring-site estimates with no corroborating status disclosure, so treat the counts as unverified.
Alternatives to ElevenLabs
GeminiGoogle's AI assistant for writing, research, coding and everyday questions, free with a Google Account and upgradable through Google AI Plus, Pro and Ultra plans.
Z.aiZhipu AI's international chat and agent platform on its GLM models, with free web chat, credit-metered coding plans for Claude Code, Cline and OpenCode, and a per-token API.
GrokxAI's conversational assistant with real-time web and X search, voice mode, and Grok Imagine image and video generation on web, iOS and Android.
JuicyChat AIAdult character chat with community-made and custom personas, plus coin-metered images and voice on web, iOS and Android.
MeshyBrowser workspace that turns text prompts and reference images into textured, riggable 3D models, with printability checks, a REST API and an MCP server.
Base44Wix-owned vibe coding platform that turns a plain-language brief into a hosted web app with data, logins, permissions and backend logic.
Frequently asked questions
Company & links
- Company
- ElevenLabs
- Founded
- 2022
- Founders
- Mati Staniszewski, Piotr Dąbkowski
- Leadership
- Mati Staniszewski is co-founder and CEO, and Piotr Dąbkowski is co-founder and CTO. They started the company in London in 2022, motivated by the quality and accessibility of synthetic speech at the time. Andrew Reed joined the board in February 2026 alongside the Series D. The full executive roster was not documented in the sources we checked.
- Funding
- ElevenLabs is private and independent. Its February 4, 2026 announcement reports $781 million raised across five rounds since 2022, though a third-party snapshot puts the total at $881 million without a round breakdown, so the total is unreconciled. The disclosed rounds include a $500 million Series D in February 2026 at an $11 billion valuation, led by Sequoia Capital with Andreessen Horowitz and ICONIQ participating heavily; a $180 million Series C in January 2025; and a $100 million employee share sale in September 2025 at a reported $6.6 billion valuation, which was a secondary transaction rather than a primary round.
- Scale
- The company says it closed 2025 above $330 million in annual recurring revenue, which is self-reported and not independently verified. A third-party snapshot reports 754 workers in 2026, also uncorroborated. Its own account describes an international footprint across London, New York, San Francisco, Warsaw, Dublin, Tokyo, Seoul, Singapore, Bengaluru, Sydney, São Paulo, Berlin, Paris and Mexico City, with London named as headquarters in one profile and a New York address in another, so the headquarters evidence conflicts.
Try ElevenLabs on work you already know
A task you’ve done before is the fastest way to judge whether ElevenLabs fits your workflow and budget. Check the current plan terms on the official site before you subscribe.