ElevenLabs pricing (2026): plans, credits & voice cloning

Plan access was checked September 13, 2026 against ElevenLabs' voice workspace guide, Professional Voice Cloning documentation and voice-cloning plan comparison. Prices are USD before tax; API and usage-based products have separate rates.

For an existing video that needs another language, compare AI dubbing tools. For changing an existing voice performance, compare AI voice changers. These are different jobs from generating speech from text.

This pricing guide uses current ElevenLabs plan, credit, and feature documentation and separates subscription allowances from usage-based API billing. It is a price reference rather than a voice-quality ranking; verify the final account terms and exact model route before purchase.

Current subscription prices

ElevenLabs pricing page, captured September 25, 2026. Billing selection, currency and any promotion are shown in the screenshot; recheck the live page and checkout before buying. View official source

Check the selected product and billing period. API pricing and usage-based products can use different units and rates.

How do credits translate into audio?

There is no single accurate credits-to-minutes conversion for every ElevenLabs workflow. Consumption depends on the selected model, feature, and input. Text-to-speech, voice conversion, dubbing, and conversational agents are different operations.

Budget dubbing separately from narration

All products draw from the shared allowance. The current pricing FAQ lists these approximate dubbing costs; watermark choice and workflow materially change the budget.

A 20-minute automatic dub therefore needs roughly 40,000 credits with a watermark or 60,000 without, before other work or revisions. Both exceed Starter's listed 30,000-credit allowance. This is arithmetic from published rates, not a measured project. Reserve the remaining pool for narration and other tools rather than counting the same credits twice.

Estimate your own monthly cost

Generate a representative passage containing your usual names, punctuation, and speaking style. Record the feature, model, charged usage, audio duration, and corrections. Scale that observed job cost by your planned workload, including alternate takes.

For example, if your own completed narration consumes 2,000 credits including revisions, ten comparable narrations would require about 20,000 credits. This is an illustrative calculation, not a fixed ElevenLabs rate or a promise that every script costs the same.

When a retry uses more credits

Credits are consumed when you generate, not when you download. ElevenLabs' regeneration rules describe limited free retries for eligible website Text to Speech and Voice Changer requests: retain the same input, voice and model, retry within two hours, and stay on the page without refreshing. Model exceptions apply; use the interface's eligible Regenerate action rather than assuming another Generate click is free. API generations are excluded. Budget paid attempts unless eligibility is shown.

Which plan should you start with?

Use the free tier to assess pronunciation and the editor. Before commercial publication or cloning, confirm that the chosen plan includes the required rights and feature. Choose the lowest qualifying plan with enough capacity; upgrade for a documented need rather than a general label such as “professional.”

Once your permitted audio is ready, choose the video workflow by input: use Talking Photo for a still portrait, Lip Sync for an existing speaking clip, or Audio-to-Video for newly generated visuals. Animation is a separate job and may add cost.

Which ElevenLabs plans include voice cloning?

Instant Voice Cloning is available on Starter and above. Professional Voice Cloning is available on Creator and above. Current base PVC slots are one on Creator and Pro, three on Scale, ten on Business, and custom on Enterprise. Free and Starter have no PVC slot. Professional Voice Clones require verification and can only be created for your own voice; a verified owner can share a clone privately.

Confirm voice eligibility and the provider's verification requirements before cloning. Include the usage allowance, extra credits, and seats in your plan decision.

Instant or Professional cloning: prepare the right recording

The Instant Voice Cloning guide recommends one consistent speaker without background noise or reverb. Adding unrelated, noisy recordings merely to increase duration can make the input less useful. These are preparation recommendations, not a guarantee that every voice will clone accurately.

The Professional Voice Cloning guide requires ownership verification and a larger recording set. Buying Creator does not make an unprepared voice immediately usable. Record the intended speaking style and keep the clean source files; check the clone’s actual readiness before booking production around it.

Can you export an ElevenLabs voice clone?

The Professional Voice Cloning FAQ says the clone itself cannot be downloaded as a standalone file. It remains in the ElevenLabs account. API access can use an account’s voices in another service; that is different from owning portable model weights. Keep your authorized source recordings if future recreation or provider migration matters.

Can you use ElevenLabs commercially?

For the speech workflow covered here, paid-plan generation carries commercial-use rights. ElevenLabs' billing documentation says speech generated on paid plans receives commercial-use rights. Free-plan speech is limited to non-commercial use with attribution. Those plan rights do not grant permission to clone another person's voice or use protected scripts, music, or recordings; confirm the rights to every input separately.

Does annual billing fit your workload?

At the published annual formula of ten monthly payments, Starter is $60 per year, Creator $220 and Pro $990, before taxes or promotions. Annual payment is upfront; ElevenLabs' annual-plan guide says self-serve credits still reset monthly, with eligible rollover. You do not receive twelve monthly allowances to spend immediately. Annual subscriptions renew until canceled; Enterprise allocation may be monthly or upfront under its agreement. Start monthly for a short project; consider annual only when continued use justifies the commitment.

Do unused ElevenLabs credits roll over?

On self-serve paid plans, credits can roll over for up to two months, capped at twice the monthly quota carried forward. That means the account balance can reach at most three times the monthly allowance after a renewal. The current pricing FAQ says rollover requires an active paid subscription; it does not apply to Free, and unused paid credits are forfeited when a downgrade or cancellation takes effect. Pay-as-you-go credits follow separate terms.

Pay As You Go: top-ups, expiry and spending controls

ElevenLabs’ current Pay As You Go guide describes prepaid top-ups for new self-serve customers. Existing legacy overage customers and Enterprise contracts retain separate billing. Subscription credits are used first; PAYG funds expire 12 months after purchase.

In Developer settings → Top Up, choose manual funding or Auto Top Up. The optional monthly spend cap is off by default. A separate plan-based monthly purchase cap applies without two-factor authentication. A PAYG balance does not itself raise your voice-slot allowance or change your subscription tier.

PAYG keeps the unused dollar value across plan changes, but the displayed credit count can change with the plan’s conversion rate. For intermittent work, compare the exact website/API rate and the features you need before funding a balance. Keep automatic funding off when a fixed project budget matters more than uninterrupted generation; review the offered cap before enabling it.

What can you keep using after cancellation?

For speech generated during a paid subscription, ElevenLabs’ account documentation says the commercial license continues after the subscription ends. That differs from continued cloud storage: it does not guarantee permanent access to generated files on the free tier. Download approved audio before ending a project.

A Professional Voice Clone remains in the account after a downgrade, but the clone-access FAQ says Starter and Free cannot use it; Creator or above is needed again. Keep licensed finished audio, source recordings, and the generation date together. Keeping a finished file does not give you portable clone weights or permission to clone somebody else.

Are ElevenLabs app credits and API costs the same?

Do not assume one universal conversion. ElevenLabs' API pricing page lists rates by product and billing model: text to speech can be metered per character, while transcription, voice processing, music, sound effects, and dubbing use different units. Subscription credits can be shared across products, but the charge depends on the product, model, plan, and whether the work runs through the website or API. Price the exact endpoint and expected retries before choosing a plan or pay-as-you-go usage.

Turn an ElevenLabs plan into a production budget

Estimate approved audio, not submitted characters. Start with script length, languages, voices, retakes, alternate reads, dubbing minutes, projects and expected monthly revisions. Use the live estimator or documented unit for the exact model and feature. Voice generation, dubbing, sound effects and other tools can consume allowances differently.

Create one representative sample before an annual commitment: include a difficult name, a number, an acronym, an emotional turn and the target language. Count every take required to pass pronunciation, pacing, tone, noise and client review. Divide plan and overage cost by approved minutes.

Keep consent and rights separate from pricing. A paid plan does not authorize cloning another person. Save the speaker’s permission, approved use, source recording and final script. Confirm the exact plan’s commercial terms and any verification requirements for the voice workflow.

When the voice must become video

If audio is one input to a talking portrait or existing performance, price the complete deliverable. Exporting speech, transferring it, fixing timing and regenerating a video can cost more operator time than the voice itself.

For a portrait, test Magic Hour Talking Photo. For existing footage, use Lip Sync. Evaluate pronunciation, mouth timing, facial stability, resolution and accepted-output cost together.