ElevenLabs API Pricing (2026): Cost per 1,000 Characters with Eleven v4, v4 Turbo, v3 and Flash, What Each Plan Includes, and Pay-As-You-Go

Short answer: the ElevenLabs API bills text to speech by the character, in US dollars, and there are only two prices. US$0.08 per 1,000 characters for the quality models (the new Eleven v4, Eleven v3 and Multilingual v2), and US$0.04 per 1,000 characters for the real-time models (Eleven v4 Turbo, v3 Conversational and Flash). ElevenLabs counts 1,000 characters as about one minute of speech, so an hour of v4 audio costs about US$4.80 and an hour of v4 Turbo or Flash about US$2.40 (our math). The per-character price is the same on every plan. A bigger plan buys more included characters, more parallel requests and plan features, not a cheaper rate. With no subscription, you can prepay from US$5 with pay-as-you-go.

Text to speech API price per model

From the model cards on elevenlabs.io/pricing/api (ElevenAPI tab) and the character limits in the developer docs (Models page). One request can hold up to the character limit; for longer text you split it into several requests.

The older eleven_turbo_v2_5 and eleven_turbo_v2 models are now listed as deprecated: “We recommend using the Flash models over Turbo models in all use cases.” On the pricing page they share the “Flash / Turbo” row at US$0.04.

Get an ElevenLabs API key (PartnerStack referral)

What is new in Eleven v4 (and is it worth the price)?

• Same price as v3. v4 costs US$0.08 per 1,000 characters, like v3 and Multilingual v2, and ElevenLabs says “Eleven v4 is a net upgrade over Eleven v3, delivering better results in almost every case.” So for most new projects, there is no price reason to stay on v3.
• Two variants. eleven_v4 is “our highest-quality model” for content, audiobooks and character voiceovers. eleven_v4_turbo is half the price and “purpose-built for real-time use cases” such as voice agents, with a median inference latency of about 100 ms.
• Endpoints. The help center says you can use eleven_v4 with the standard Create speech and Stream speech endpoints; the Models page also lists v4 for the Text to Dialogue API, and v4 Turbo for the Text to Dialogue websocket.
• Fewer knobs. “Both variants use two voice settings: Stability and Similarity.” “Style and Speed sliders are not available in Eleven v4, and SSML is not supported.” You direct delivery with audio tags such as [whispering] instead. Professional Voice Clones get an extra Clone Strength slider.
• Accents change across languages. When you generate in a language other than the voice’s own, v4 speaks it with a native accent: “if you clone a voice from a Korean speaker and generate English, Eleven v4 will produce natural, fluent English rather than English spoken with a Korean accent.” If your product relies on a voice keeping its accent, test this before you switch.

What each plan includes for the API

The “Model pricing” table on the same page shows how many characters each plan’s monthly fee covers. Concurrency (how many requests run at once) comes from the docs’ Concurrency and priority table.

The key point (our math): the included characters are worth almost exactly the plan price at list rate. 275,000 characters × US$0.08 per 1,000 = US$22, which is the Creator price; Pro’s 1,238,000 characters come to US$99.04, Business’s to US$990. So a subscription is effectively prepaid usage at the normal rate. What you get on top is the plan itself: a commercial license on paid plans, more voice slots, higher concurrency and priority, and on Pro and above “44.1kHz PCM audio output via API”. Yearly billing is the monthly price × 10 (Creator works out to US$18.33 a month, Pro to US$82.50), which is the only real per-character discount on self-serve plans.

The Starter US$1 first month is a limited offer: the main pricing page says it runs “Until Nov 1”. After that, Starter is US$6.

Pay-as-you-go: using the API without a subscription

• Prepaid balance. You add funds under Developers > Top Up. The minimum top-up is US$5, and “PAYG is a prepaid model with no overage charges.” If your balance hits $0 and Auto Top Up is off (or fails), “your service will pause immediately.”
• Auto Top Up recharges a default of US$20 when your balance falls below US$10 (both adjustable).
• Order of use. If you also have a subscription, “your subscription plan’s included credits are consumed first”, and then the balance takes over. This is how you go past your plan: PAYG “replaces the previous usage based billing (overage) model for new self-serve customers.”
• Expiry. “PAYG credits expire 12 months after purchase.” Separately, if you downgrade or cancel a plan, “any unused paid credits will expire” at the end of the billing cycle.
• Monthly cap without 2FA. Without two-factor authentication you can top up at most US$100 a month on Free, Starter and Creator, US$200 on Pro, US$600 on Scale and US$2,000 on Business. “Enable 2FA to remove the cap.”
• Limits still come from your plan. A Free account with a balance runs at Starter-level concurrency (3 requests) and priority, but “voice slot limits and other plan-level restrictions (e.g., custom voice limits, API access tiers) come from your subscription plan, not your PAYG balance.”

Other ElevenLabs API prices

Burst pricing (US$0.16 a minute) “applies to connected audio above your concurrency limit” for the Realtime models and Speech Engine. You can follow your monthly usage in the analytics tab of the Developers dashboard.

What real projects cost (our math)

These use ElevenLabs’ list prices and its “1,000 characters ≈ 1 minute” estimate. Your real character count depends on your text.

• A 100,000-character month (the calculator’s default): US$8 on pure pay-as-you-go at US$0.08, or Starter (US$6 for 75,000) plus US$2 of top-up, also US$8. The calculator itself recommends Creator for this, which leaves 175,000 characters spare; pick Creator only if you want the headroom or its features.
• One million characters (about 16.7 hours of audio): US$80 on v4, US$40 on v4 Turbo or Flash.
• An 8-hour audiobook on v4: about 480,000 characters, roughly US$38.40 at list rate. It fits inside one month of Pro (1,238,000 characters).
• A voice agent speaking 500 minutes a month: about US$40 on v4 or US$20 on v4 Turbo for the speech alone. LLM and telephony costs are separate.

Which model and plan to pick

Start with ElevenLabs (PartnerStack referral)

More from ElevenLabs on SiftGuide

Using the web app rather than the API? Credits work differently there; see the ElevenLabs pricing guide. For long-form narration in the editor, see ElevenLabs Studio pricing. Building an agent for your site? See how to add an ElevenLabs AI voice agent to your website.

FAQ

How much does the ElevenLabs API cost?
Text to speech is US$0.08 per 1,000 characters on Eleven v4, v3 and Multilingual v2, and US$0.04 on v4 Turbo, v3 Conversational and Flash. It is billed in US dollars, not credits, and prices exclude taxes.

Is the ElevenLabs API free?
The Free plan includes 10,000 characters a month on the US$0.08 models (20,000 on the US$0.04 models), plus 4.5 hours of Scribe v2 transcription. Free use is limited to non-commercial purposes under the Terms.

How much does Eleven v4 cost?
US$0.08 per 1,000 characters, the same as v3. Eleven v4 Turbo is US$0.04.

Do bigger plans get a cheaper rate per character?
No. The price per 1,000 characters is the same on every plan; bigger plans include more characters and higher concurrency. Yearly billing (10 × the monthly price) is the main discount.

What happens when I run out of included characters?
New self-serve accounts top up a prepaid pay-as-you-go balance (minimum US$5, optional Auto Top Up). Your plan’s credits are used first; if the balance reaches $0, service pauses.

Can I use my cloned voices through the API?
Yes. You can reference any voice in your library by ID, including professional, cloned and designed voices.

Sources

All checked on October 23, 2026:

• ElevenLabs API pricing (model cards, Model pricing table, calculator, FAQ)
• ElevenLabs pricing (plan cards, annual prices)
• ElevenLabs Terms of Service
• Models (model IDs, character limits, concurrency and priority)
• Eleven v4
• Help center: What is Eleven v4?
• Pay As You Go