ElevenLabs Review: Models, Pricing & How to Start
ElevenLabs offers AI voice generation, voice cloning, dubbing and transcription. We tested Eleven v4 in English: 18 of 18 checkpoints correct.
👁 148 reads
ElevenLabs is a service for text-to-speech generation, voice cloning, dubbing and transcription, founded in 2022. On September 28, 2026, its new model Eleven v4 took first place in the Artificial Analysis voice ranking. We’ve gathered everything you need to get started: models, pricing, first steps, features, API and alternatives, with results from our tests.
Summary
ElevenLabs has the highest-rated speech synthesis on the market (1319 Elo points in Artificial Analysis), and in our English test Eleven v4 read all numbers, dates and abbreviations correctly (18 of 18 checkpoints). The free plan gives you 10,000 credits, about 15 minutes of speech in v4, without commercial use rights. A commercial license comes with Starter at $6/month, and Creator at $22 adds professional voice cloning. Who it’s for: video creators, course producers, podcasters and voice app developers. Who it’s not for: if you just want to listen to articles read aloud, free system tools are enough.
01Eleven v4 Review: Numbers, Dates & Emotion Tags in EnglishEleven v4 tops the Artificial Analysis TTS leaderboard with 1319 Elo. We tested it in English: 18/18 numbers and abbreviations correct, 0.25s to first audio in Turbo.Read →
02Best AI Voice Generator for Polish: 14 TTS Models RankedWe tested 14 AI text-to-speech models on complex Polish sentences, dates, and abbreviations. ElevenLabs v4 took the crown alongside Gemini 3.1 Flash TTS.Read →
03ElevenLabs Dubbing v1 vs v2: Test on a 24-Second ClipElevenLabs Dubbing v1 vs v2 tested on a 24-second clip: time, credits, voice similarity, and one failure that still cost credits.Read →
04ElevenLabs Voice Cloning: clones from 10 s, 30 s and 2 minWe cloned a synthetic voice from 10 s, 30 s and 2 minutes of audio. The 10-second clone reached 0.992 similarity, the 2-minute clone 0.996.Read →
05Eleven Music: music_v2_5 vs music_v1 API Default TestWe tested Eleven Music's music_v2_5 against the API default music_v1 with English lyrics. Both models sang clearly and hit the requested genres.Read →
06ElevenLabs Sound Effects: Prompt Influence 0.3 vs 1.0 TestWe tested ElevenLabs Sound Effects with five English prompts at prompt influence 0.3 and 1.0. Both settings hit the target 5 of 5 times.Read →
07ElevenLabs Scribe: English transcription test with hard namesElevenLabs Scribe v2 transcribed a 59-second English interview with tricky names and 12 numbers in 3.6 s with 2 automatic differences, correctly attributed 6 of 6 turns.Read →
08ElevenLabs Voice Changer: AI Voice Swap Test & ResultsVoice Changer keeps your timing and melody while swapping the voice. We tested it on clean and noisy recordings - here's what survived and what didn't.Read →
09Voice Isolator: removing noise from audio with ElevenLabsVoice Isolator from ElevenLabs removes background noise from recordings. We tested it with cafe chatter, street noise and fan hum to measure how much cleaner it gets.Read →
10ElevenLabs Agents Voicebot Test: Dental Reception LineWe built a dental reception voicebot with ElevenLabs Agents and tested it with real-time voice calls: response time, speech recognition, cost per minute.Read →What is ElevenLabs
ElevenLabs started with one thing: turning text into speech that doesn’t sound like a train station announcement. Today it’s a suite of audio tools: text-to-speech narration, voice cloning and design, video dubbing into other languages, transcription, sound effects, music, and voice agents that conduct phone conversations. Everything works in the browser (ElevenCreative app), on mobile, and via API.
The app interface is in English, and it generates speech in 85 languages including English, Spanish, French, German, Italian, Portuguese, Dutch, Polish, Arabic, Hindi, Japanese, Chinese, Korean and many others.
ElevenLabs Models
The model determines quality, speed and cost. List from the API as of September 28, 2026, and results from our English test of numbers and abbreviations (18 checkpoints, native voice Jon):
| Model | Languages | Credits per char | What for | Our test (of 18) |
|---|---|---|---|---|
| Eleven v4 | 85 | 1 | narration, audiobooks, ads, emotion tags | 18 |
| Eleven v4 Turbo | 85 | 0.5 | voice assistants, live conversation | 18 |
| Eleven v3 | 74 | 1 | previous generation with tags | 18 |
| Multilingual v2 | 29 | 1 | long, stable recordings | 10 |
| Flash v2.5 | 32 | 0.5 | lowest latency (0.23 s) | 11 (10 of 11 on obstacle course with normalization) |
For English text, choose v4 or v4 Turbo. Older models sometimes read abbreviations literally and confused amounts. We tested all models in English on September 29, 2026, with native voices Jon (male) and Hope (female), plus the built-in Alice voice.
ElevenLabs Pricing
Prices from the official pricing page, checked September 28, 2026, excluding VAT (ElevenLabs lists prices without taxes). Minutes calculated from our recordings: one minute of English speech is about 660 characters, or 660 credits in v4.
| Plan | Monthly | Annual billing | Credits | Speech in v4 | Key features |
|---|---|---|---|---|---|
| Free | $0 | - | 10,000 | ~15 min | no commercial license |
| Starter | $6 | $5 | 30,000 | ~45 min | commercial license, instant voice clone, dubbing |
| Creator | $22, 1st month $11 | $18.33 | 121,000 | ~3 h | professional voice clone |
| Pro | $99 | $82.50 | 600,000 | ~15 h | 44.1 kHz PCM quality in API |
| Scale | $299 | $249.17 | 1.8M | ~45 h | 3 seats, 3 professional clones |
| Business | $990 | $825 | 6M | ~150 h | 10 seats, 10 clones |
| Enterprise | custom quote | - | - | - | SLA agreements, SSO |
In v4 Turbo and Flash v2.5 you get twice as many minutes because they use half a credit per character. Unused credits roll over to the following months, but for a maximum of two months.
How to Start with ElevenLabs
- Create an account at ElevenLabs (email or Google account is enough) and go to Text to Speech.
- Paste your text. Short sentences and commas at breathing points improve the result in every model.
- Choose a voice: in Voice Library you can filter by language. Popular English voices include Jon, Hope, Alice, and many others.
- Select the Eleven v4 model and click Generate. Download the MP3 file or generate again if something sounds wrong.
In v4 you can add emotion tags in brackets, such as [whispers] or [sighs], which the model performs instead of reading aloud.
ElevenLabs Features
Text to Speech
The core feature: you paste text, choose a voice and model, and get a recording. In v4 you can add emotion tags in brackets, like [whispers] or [sighs], which the model performs instead of reading. We tested this in English on September 29, 2026.
Voice Cloning
Instant Voice Cloning, available from the Starter plan, according to ElevenLabs is created in v4 from 10 seconds of recording. Professional Voice Cloning, available from the Creator plan, requires at least 30 minutes of recordings and verification that the voice belongs to you. You may only clone your own voice or the voice of a person who has given explicit consent; ElevenLabs blocks voices of well-known individuals and prohibits impersonating politicians even with consent.
Dubbing
Dubbing translates a video into one of over 90 languages, preserves the voice timbre of speakers, and supports up to 32 speakers in a single file. In the API it costs $2.20 per minute in v2. The voice in dubbing is also a specific person’s voice, so before publishing you need consent from everyone whose timbre will be reproduced in another language.
Speech to Text (Scribe)
Transcription of recordings to text, in the API for $0.22 per hour of recording. We used Scribe v2 as one of three judges in our narrator test: it transcribed a 32-second recording in 2.8 s and recognized English with 100% confidence.
Sound Effects and Music
From a text description, ElevenLabs generates sound effects (footsteps, rain, doors) and music tracks that can be used commercially on paid plans.
Voice Changer and Voice Isolator
Voice Changer transforms your voice in a recording to another while preserving intonation, and Voice Isolator removes noise and background from a recording. In the API both features cost $0.12 per minute.
Studio and Audiobooks
Studio is an editor for longer projects: book chapters, courses and podcasts, with the ability to assign different voices to characters.
Voice Agents
ElevenAgents is a platform for building assistants that talk to customers by phone or on a website. For such conversations, v4 Turbo is useful - in our measurement it started speaking after 0.25 s. ElevenLabs rules require informing the caller that they are talking to AI.
ElevenLabs API
The API is billed separately from app plans. Pricing as of September 28, 2026, per 1,000 characters, excluding VAT:
| Model | Price per 1,000 chars | Per million chars | Time to first audio (our measurement) |
|---|---|---|---|
| Eleven v4 | $0.08, promo until Oct 12: $0.022 | $80 | 1.42 s |
| Eleven v4 Turbo | $0.04, promo until Oct 12: $0.011 | $40 | 0.251 s |
| Eleven v3 | $0.08 | $80 | 0.859 s |
| Flash v2.5 | $0.04 | $40 | 0.229 s |
You create a key in account settings, and a call is one POST request with text and voice identifier. For English text, add language_code: "en", and for Flash add apply_text_normalization: "on" - without it, in our test it confused numbers. Time-to-first-audio measured in streaming mode, median of three attempts.
Alternatives to ElevenLabs
- Google Gemini TTS (3.8 Flash, 3.8 Flash-Lite, 3.1 Flash): $11-18.3 per million characters according to Artificial Analysis. We could not test it on September 29, 2026, due to account limitations.
- Cartesia Sonic 3.6: 2nd place in the Artificial Analysis ranking, $49 per million characters, plans from $5. We did not test it because we had no API access.
- OpenAI GPT Audio: good English pronunciation, but as a chat model it can treat text as a command and refuse to read it. In our test, GPT Audio refused to read the brand-name sentence and answered it like a request, so every recording from a chat model has to be checked.
- Free and local: system voices and open models are available in many operating systems and frameworks.
Frequently Asked Questions
Is ElevenLabs free?
There is a free plan with 10,000 credits per month, about 15 minutes of speech in Eleven v4. Recordings from it may not be used commercially, and publication requires ElevenLabs attribution.
How much does ElevenLabs cost?
Starter costs $6 per month (excl. VAT), Creator $22, Pro $99. With annual billing, prices are about 17% lower.
Does ElevenLabs have an API?
Yes. Eleven v4 costs $0.08 per 1,000 characters in the API, with a promotional price of $0.022 until October 12, 2026. The API also supports transcription, dubbing, voice cloning and voice agents.
Can you clone a voice in ElevenLabs?
Yes, from the Starter plan (instant clone) and Creator (professional clone with verification). You may only clone your own voice or the voice of a person who has given consent.
Sources and verification date
- ElevenLabs plan pricing
- API pricing
- model documentation
- Eleven v4 launch
- dubbing
- professional voice cloning
- use policy
- safety. Ranking: Artificial Analysis. Alternatives: Gemini API
- Cartesia. Model list
- languages and English voices: ElevenLabs API
- read September 28
- 2026. English test conducted September 29
- 2026. Verified September 28-29
- 2026
ElevenLabs Voice Cloning: clones from 10 s, 30 s and 2 minAI Tools
ElevenLabs Use Cases: Voiceovers, Voice Cloning & AudiobooksGuides
Voice Isolator: removing noise from audio with ElevenLabsAI ToolsThe link to ElevenLabs is a partner link: if you buy through it, QUANTSLANT earns a commission and you pay the same price. It has no influence on the content, including where the coverage is critical.
More on this from QUANTSLANT
news·OpenAI Agents API: Computer Use, Decisions API & PricingOpenAI's Agents API now includes computer use: agents open browsers in the cloud and click through sites. Plus Decisions API on Luna and pricing breakdown.