ElevenLabs in 2026: Pricing, Free Plan, Voice Cloning and Dubbing

ElevenLabs is an AI voice platform: it turns text into realistic speech, clones voices, dubs video into 90+ languages, composes music and runs voice agents. It’s built for video creators, audiobook narrators, localization teams and developers. The free plan covers about 10 minutes of speech a month for non-commercial use, and paid plans start at $6 a month plus tax (pricing, checked Sept. 25, 2026).

This guide covers documented pricing, quotas and features. Voice models were rechecked on September 28, 2026; dubbing workflows and billing on September 30. Other details retain their individual verification dates.

What Is ElevenLabs in 2026?

The company now groups its products into three areas (homepage, checked Sept. 25, 2026):

• ElevenCreative, for making content: text to speech, Studio for long-form projects, voice cloning and voice design, dubbing, music, sound effects, voice isolation, and an image and video module.
• ElevenAgents, for building voice agents (customer support, appointment booking, lead qualification) on a website, in an app or over the phone.
• ElevenAPI, for developers who build these features into their own products.

All three run on the same account and, for the most part, draw on the same monthly credit balance.

What’s Changed Since 2025

• Eleven v4 and v4 Turbo launched on September 28, 2026 across ElevenCreative, ElevenAgents and the API. v4 supports 90+ languages and both instant and professional voice clones; v4 Turbo targets real-time conversations (official announcement, checked Sept. 28, 2026).

• Eleven v3 came out of alpha on Feb. 2, 2026. ElevenLabs reports 68% fewer errors when reading numbers, formulas, URLs or phone numbers aloud, based on its own benchmark (announcement).

• Dubbing v2 (May 28, 2026) carries the speaker’s intonation and emotion across languages, in 90+ languages (announcement). The documentation still labels it alpha as of Sept. 30, 2026.

• Music v2 (May 26, 2026), followed by Music v2.5 (Sept. 11, 2026), now the default music model (Music v2.5 announcement).

• Image & Video (in beta since Nov. 17, 2025): generate images and video with third-party models such as Veo, Sora, Kling, Wan or Seedance, then add voiceover, music and sound effects (announcement).

• Plans are now measured in credits rather than characters, and both prices and quotas have changed.

How Much Does ElevenLabs Cost?

Prices are in U.S. dollars and exclude sales tax. Paying yearly works out to 12 months for the price of 10 (ElevenLabs pricing, checked Sept. 25, 2026).

Billed yearly, Starter comes to $5 a month and Creator to $18.33 a month.

How Many Credits Does Each Feature Use?

The features below draw from your credit balance. The pricing FAQ lists these approximate costs; its four dubbing rates apply to the legacy v1 workflow and Dubbing Studio, not Dubbing v2 (distinction rechecked Sept. 30, 2026 against the dubbing documentation).

Budget Dubbing v2 separately. API pricing lists $2.20 per minute, excluding taxes, for this model (checked Sept. 30, 2026). The dubbing billing guide specifies that each target language is charged based on source duration. That API rate does not establish your in-app workflow’s cost: the app displays a total before confirmation based on model, duration and languages. Watermarking depends on whether you use a free or paid tier; v2 does not offer a watermark discount. The minute estimates in this table therefore do not apply to v2.

Unused credits on a paid plan roll over for up to two months as long as the subscription stays active; they’re lost if you cancel or downgrade.

Which Plan Should You Pick?

• Trying out the voices: the free plan is enough.
• Publishing or monetizing videos, cloning your own voice: Starter, at $6 a month.
• Audiobooks or regular voiceover, a high-fidelity cloned voice: Creator, with 121 minutes a month and Professional Voice Cloning.
• API integration or voice agents in production: Pro and up, depending on volume.

Is ElevenLabs Free?

Yes, for trying it out: the free plan gives you 10,000 credits a month, about ten minutes of speech, with no rollover. You get Eleven v3, Voice Design, music and 15 minutes of voice agent calls, but no voice cloning (pricing, checked Sept. 25, 2026).

Two rules to know before you publish anything (publishing rules, checked Sept. 25, 2026):

• No commercial use: a monetized video or an ad needs a paid plan.
• Attribution required: published content must include “elevenlabs.io” or “11.ai” in its title.

For a YouTube channel or client work, Starter at $6 a month is the real entry point.

Which Voice Models and Languages Are Available?

ElevenLabs offers several text-to-speech models, each suited to a different job (models documentation, checked Sept. 28, 2026):

The older Turbo v2 and v2.5 models were replaced by Flash; the new v4 Turbo is a separate model. Language coverage depends on the selected model.

For changes from v3, checks before migrating existing voices and a worked API budget, read the Eleven v4 and v4 Turbo launch report.

Directing Eleven v3 With Audio Tags

Eleven v3 takes audio tags in square brackets, placed right in the text, to make a voice whisper, laugh, sound sarcastic or curious (best practices guide). For example:

[whispers] I really shouldn't tell you this… [laughs] but the project is already done.

Eleven v3 doesn’t support SSML <break> tags. The docs recommend ellipses and punctuation to control pacing instead. For a step-by-step walkthrough, see our tutorial on creating realistic AI voices with ElevenLabs.

How Does ElevenLabs Voice Cloning Work?

ElevenLabs offers two levels of cloning (voice cloning docs, checked Sept. 25, 2026):

If you don’t want to clone anyone, Voice Design creates a voice from a written description, even on the free plan.

The usage policy prohibits replicating another person’s voice without their consent or a legal right. For a brand voice, get written consent from the person you record.

Dubbing, Audiobooks, Transcription: What Else Can It Do?

Automatic Dubbing v2 Alpha translates video or audio into 90+ languages while preserving speakers’ vocal characteristics and background audio. The app accepts up to 32 speakers and 180 minutes. There is no in-app editor; transcript editing and regeneration through the API require Enterprise (dubbing documentation, checked Sept. 30, 2026).

Dubbing Studio is a separate workflow using v1. It supports transcript corrections, speaker reassignment and per-clip audio regeneration, with a 45-minute, 1 GB upload limit. The Studio guide now describes it as being in maintenance mode with critical fixes only. The pricing page, checked Sept. 30, 2026, includes access from Starter. Do not choose v2 expecting to continue editing its output in Studio.

For source preparation, translation checks and budgeting, follow the guide to translating a video while keeping your voice.

Studio is built for long-form work: audiobooks, podcasts and video voiceovers, such as the narration for an educational video. It imports a document (EPUB, PDF, DOCX, TXT, HTML or a web page), detects its chapters and lets you assign multiple voices (Studio docs, checked Sept. 25, 2026). The free plan allows 3 Studio projects, Starter 20.

Other tools:

• Speech to Text (Scribe v2 model): transcription in 90+ languages, batch or real time.
• Voice Changer: re-voices your recording with a different voice while keeping your delivery.
• Voice Isolator: strips background noise from a recording.
• Sound Effects: generates sound effects from a text description.

Music

Music v2.5 generates tracks from 3 seconds to 5 minutes long, with or without lyrics, in several languages (music docs, checked Sept. 25, 2026). ElevenLabs positions it as cleared for commercial use, built with licensing deals (Believe for Music v2, then a multi-year agreement with Universal Music Group mentioned in the Music v2.5 announcement). Commercial use of music starts with the Starter plan. On the free plan, any track you publish has to credit “Eleven Music” (publishing rules, checked Sept. 25, 2026). Music is also available in a separate app, ElevenMusic.

If you just need instrumental background music, Soundraw is a simpler option.

Image and Video

This module is still in beta. It gives you image models (Nano Banana, Flux Kontext, GPT Image, Seedream) and video models (Veo, Sora, Kling, Wan, Seedance), plus lip sync and audio finishing. Sora is still listed on ElevenLabs’ video pages, but OpenAI shut down its API on Sept. 24, 2026, so that access is now uncertain (checked Sept. 26, 2026). One catch: the commercial license on paid plans doesn’t cover beta services. To compare those video models, see our top 10 AI video tools.

Voice Agents and API: What Do They Cost?

ElevenAgents builds agents that listen and talk back, with a knowledge base, a workflow builder, an embeddable website widget and phone integration. Every plan includes call minutes (ElevenAgents pricing, checked Sept. 25, 2026):

Extra minutes cost $0.08 each. The language model (Gemini, Claude, GPT and others) is billed on top depending on which one you pick, and telephony is passed through at cost.

The API lets developers integrate features into an application; billing depends on the feature and model. Flash v2.5 costs half as many credits per character as the high-quality models, and 44.1 kHz PCM output is limited to Pro and above. The official documentation covers the API and agents.

What Are ElevenLabs’ Limitations?

• Budget for the workflow: Dubbing Studio v1 uses more credits per minute than voiceover; v2 needs a separate cost check. Include additional languages and corrections before choosing a plan.
• 5,000 characters per generation with Eleven v3: longer scripts have to be split, or run through Studio, which does it for you.
• Beta services aren’t covered by the commercial license, including Image & Video.
• Cloning and deepfakes: the docs only describe voice verification for Professional Voice Cloning. Disclosing that a voice is synthetic is on you, especially for anything that could mislead.
• A sprawling lineup: ElevenCreative, ElevenAgents, ElevenAPI and ElevenMusic each have their own pricing page, so it takes time to figure out what to compare.

ElevenLabs vs. Murf

Murf Studio focuses on professional voiceover: e-learning, presentations and corporate video. Its studio integrates with Canva and, on the Business plan, with PowerPoint and Google Slides. Comparison based on the pricing pages of ElevenLabs (checked Sept. 28, 2026) and Murf (Studio tab, checked Sept. 30, 2026):

Murf Studio fits script-based voiceover production with an allowance measured in hours. Its custom voice cloning remains an Enterprise add-on, while ElevenLabs offers cloning on lower-cost plans. For an existing recording, Murf Dub is a separate product: its credits are not included in the Studio comparison above.

How to Get Started With ElevenLabs

1. Create a free account on ElevenLabs (affiliate link: we earn a commission if you subscribe, at no extra cost to you).
2. Try Text to Speech with Eleven v4 on a short script, then Voice Design. The tutorial’s v3 tag examples remain useful with v3 selected.
3. Upgrade to Starter as soon as your content will be published or monetized.

FAQ

How Many Free Minutes Does ElevenLabs Give You?

About 10 minutes of speech a month (10,000 credits), for non-commercial use only and with “elevenlabs.io” or “11.ai” in the title of anything you publish. The first paid plan, Starter, costs $6 a month plus tax (checked Sept. 25, 2026 on the pricing page).

Can I Use ElevenLabs Voices Commercially?

Yes, on any paid plan, as long as you’re not using a beta service and you hold the rights to what you upload (text, cloned voice).

Which Languages Does ElevenLabs Support?

Eleven v4 covers 90+ languages, Eleven v3 covers 70+, Flash v2.5 covers 32 and Multilingual v2 covers 29. Dubbing and speech to text support 90+ languages.

How Long Does Voice Cloning Take?

Instant Voice Cloning needs only one to two minutes of audio. Professional Voice Cloning needs at least 30 minutes of audio and usually 3 to 6 hours of training.

What Happens to Unused Credits?

On a paid plan, they stay available for up to two more months, capped at twice your monthly quota. They’re lost if you cancel or downgrade. The free plan doesn’t roll anything over.