Cartesia

Text-to-speech with minutes published per tier

by Cartesia AI, Inc.

Overview

Cartesia sells voice models as one subscription whose credits are shared by the web playground and the API, so a single plan covers both typing into the browser at play.cartesia.ai and calling the text-to-speech endpoint from code (cartesia.ai/pricing and cartesia.ai/legal/terms, read 12 October 2026). The plan table publishes a monthly price at every step - Free $0, Pro $5, Startup $49, Scale $299 - and with it the conversion this kind of page usually omits: the Sonic-3.6 row states minutes included monthly as about 27 on Free, 133 on Pro, 1,667 on Startup and 10,667 on Scale, while the FAQ states that one minute of audio generation takes 750-800 credits and that one credit equals one character. The terms put the currency and the cadence in writing, saying that paid subscriptions are payable in U.S. Dollars and renew monthly from the purchase date. The hard limit to plan around is the credit pool itself: Pro carries 100K credits a month and 3 concurrent TTS requests, its overage rate is $65 per 1M credits, and with overages switched off new requests are blocked once the balance has been consumed. No region is named for any of those figures; what the page describes is buying from Cartesia directly, and the FAQ adds that a subscription can be paused or cancelled at any time.

What a finished file looks like is the half that is not settled. The API reference documents the output container as WAV, MP3 or RAW and its examples request a wav container with pcm_s16le at 44,100 Hz, so the container is stated for the API route. The playground is the surface the terms send you to for cancelling a plan, but no page read here says what it exports, so this entry records the web export format as not established rather than carrying the API's list across to a different surface.

On the three boundaries this catalogue tracks, Cartesia is unusually explicit, and one answer runs against the buyer. Commercial use has a source and a tier gate: the plan table lists a commercial use licence from Pro beside instant voice cloning, while the terms permit use of the services 'for your personal, non-commercial use only (unless commercial use is expressly permitted by your subscription tier)' and repeat that condition where outputs are concerned. Training and data use has a source and it is unfavourable: the terms state that Inputs, Outputs and interactions with the services may be used by Cartesia to train, enhance, evolve and improve its models, including for labelling, classification, content moderation and model training, with a separate section offering a form to stop selected categories of content being used in future, and the same section states that Cartesia does not claim ownership of your inputs or outputs. Whether an API is included has an answer in effect rather than in one sentence: the plan credits are described as a pool for speech-to-text and text-to-speech at their respective rates, and the playground and the API are two surfaces on that one pool, while Managed Agents are billed on their own meter at $0.06 per call minute plus $0.014 per minute for a Cartesia-provided phone number, so the credit pool does not describe voice agents.

Editor's take

Compare it on the published conversion rather than the headline price. Cartesia is one of the few speech vendors that states both what a credit buys - one character - and what a month of credits buys in minutes, so the arithmetic this brief depends on can be done from the vendor's own numbers: about 133 minutes on the $5 Pro tier. Three things decide the rest. The licence that allows commercial publishing starts at that tier, and the terms treat your material in the opposite way from a privacy-first vendor, since inputs, outputs and interactions may be used to train Cartesia's models unless the opt-out form is filed. And the export question is documented for the API rather than for the playground, so a deliverable that has to come out of a browser window without code remains unconfirmed.

Our reading of who this suits, based on the sources above. Opinion, offered to help you decide - not a lab result.

Key Features

•Sonic text-to-speech with a published conversion of 750-800 credits per minute and one credit per character
•Web playground and API drawing on one shared monthly credit pool
•Minutes included monthly stated per tier: about 27 on Free, 133 on Pro, 1,667 on Startup and 10,667 on Scale
•Instant voice cloning from Pro and professional voice cloning from Startup, with 2 cloning slots on Startup and 4 on Scale
•Commercial use licence listed from the Pro tier
•Managed Agents billed on a separate meter at $0.06 per call minute

Use Cases

→Narrating a course module or a client video from a script
→Cloning a voice and reusing it across a series of clips
→Streaming speech into an application over the API
→Trying a voice in the playground before wiring the API

Pros

✓The vendor publishes the credits-to-minutes conversion, which most speech plans leave out
✓One $5 subscription covers both the playground and the API rather than two separate meters
✓Commercial use is listed per tier on the pricing page and the terms name the same condition
✓Concurrency ceilings, overage rates and the rollover cap are all stated per tier

Cons

✕The terms let Cartesia use inputs, outputs and interactions to train and improve its models, with a request form as the remedy
✕The playground export format is not described on the pages read here, so the container is established for the API route only
✕Credits roll over only up to twice the monthly plan rate
✕Voice agents run on a separate meter, so call minutes sit outside the plan credits

Frequently Asked Questions

How much does Cartesia cost?

Free is $0 with 20K credits a month, Pro is $5 with 100K, Startup is $49 with 1.25M and Scale is $299 with 8M, each with prepaid agent dollars on the side; the terms state that paid subscriptions are payable in U.S. Dollars and renew monthly from the purchase date, and no region is named for the figures. Managed Agents are billed separately at $0.06 per call minute.

How many minutes of speech do I get?

The plan table states minutes included monthly for Sonic-3.6 as about 27 on Free, 133 on Pro, 1,667 on Startup and 10,667 on Scale, and the FAQ states that one minute of generation takes 750-800 credits with one credit equal to one character. Read those as the vendor's own conversion rather than a fixed rate for your script, since audio length depends on speaking rate and on the text itself.

Can I publish the narration commercially?

The plan table lists a commercial use licence from Pro, and the terms permit use of the services for personal, non-commercial purposes unless commercial use is expressly permitted by your subscription tier, stating the same condition for outputs. The tier active when each clip was made is therefore the fact to record, and free-tier output sits outside the published commercial licence.

Does Cartesia train on what I generate?

By default the terms say it may: inputs, outputs and interactions with the services may be used to train, enhance, evolve and improve Cartesia's models, including for labelling, classification, content moderation and model training, and a separate section offers a form to stop selected categories of content being used in future, which does not undo earlier use. The same section states that Cartesia does not claim ownership of your inputs or outputs.

Which file formats can I get?

The API reference documents the output container as WAV, MP3 or RAW, with examples requesting WAV at 44.1kHz. The pages read here do not describe an export from the playground itself, so the web export format is not established and the documented containers belong to the API route.

Do model APIs and Managed Agents share billing?

Cartesia subscriptions include monthly model credits. TTS bytes/SSE/WebSocket requests consume about one credit per character; successful requests consume credits and errors do not. Managed Agents are billed separately per minute in USD without using the model-credit balance. Overage access depends on the workspace setting.

Source: cartesia official car-api

Task guides for this product

T12 · Choose a text-to-speech route for deliverable narration

Key facts and where they come from

PriceDocumented: Pro is listed at $5 per month with 100K credits; the page names no purchase region. This is a published plan quote, not a tested account checkout. Historical scope read on 2026-10-12, not rechecked in full on 2026-10-15: Free '$0 /mo' with '20K credits / month'; Pro '$5 /mo' with '100K credits / month', 'Commercial use license' and 'Instant voice cloning'; Startup '$49 /mo' with '1.25M credits / month' and 'Pro voice cloning'; Scale '$299 /mo' with '8M credits / month'; Enterprise 'Custom'.Source: Cartesia - pricing
AvailabilityNot established: Cartesia documents APIs, subscriptions and credits, but the public pricing documentation does not establish a full country-by-country service/plan access policy. Hosting or supported speech languages are not distribution evidence.Source: Cartesia - pricing · cartesia official car-api
Commercial useDocumented: Section 4.1: 'We hereby permit you to use the Services for your personal, non-commercial use only (unless commercial use is expressly permitted by your subscription tier)'. Section 5.3(b): 'You represent and warrant that you will not monetize, make commercial use of, or otherwise use for or in connection with any commercial purposes, any of Your Outputs that you create with the Services (unless commercial use is expressly permitted by your subscription tier).Source: Cartesia - Terms of Service
Training on your dataDocumented: Section 5.3(c) 'Models': 'Inputs, Outputs, and user interactions with the Services may be used by Cartesia to train, enhance, evolve and improve its machine learning and artificial intelligence models and algorithms (collectively, "Models"), including for labeling, classification, content moderation and model training purposes.' Section 5.3(d) offers a form to opt out of training on selected categories of content, without undoing earlier use.Source: Cartesia - Terms of Service
API billingDocumented: Cartesia subscriptions include monthly model credits. TTS bytes/SSE/WebSocket requests consume about one credit per character; successful requests consume credits and errors do not. Managed Agents are billed separately per minute in USD without using the model-credit balance. Overage access depends on the workspace setting.Source: cartesia official car-api
Export formatsDocumented: The request body requires 'output_format', documented as 'WAVOutputFormat · object' with the alternatives 'WAVOutputFormat MP3OutputFormat RAWOutputFormat', and the examples send '"output_format": { "container": "wav", "encoding": "pcm_s16le", "sample_rate": 44100 }' to 'https://api.cartesia.ai/tts/bytes'. The response is 'audio/*'. The page documents the API rather than an export from the playground.Source: Cartesia - API reference, text to speech bytes
Usage limitsDocumented: Pro lists 100K credits, about 133 Sonic-3.6 minutes and three concurrent TTS requests. With overages disabled, new requests stop when the balance is exhausted; voice-agent call minutes and telephony are separately metered. Historical scope read on 2026-10-12, not rechecked in full on 2026-10-15: Free '$0 /mo' with '20K credits / month'; Pro '$5 /mo' with '100K credits / month', 'Commercial use license' and 'Instant voice cloning'; Startup '$49 /mo' with '1.25M credits / month' and 'Pro voice cloning'; Scale '$299 /mo' with '8M credits / month'; Enterprise 'Custom'. Sonic-3.6 'Minutes included monthly ~27 ~133 ~1,667 ~10,667' across Free, Pro, Startup and Scale, with 'Concurrent requests 2 3 5 15'. FAQ: 'Sonic text-to-speech: One minute of audio generation requires 750-800 credits. 1 credit equals 1 character.' / 'Your subscription will renew monthly from your purchase date.' / 'At pro tier, it's $65 per 1M credits.' / 'If your overages are off, new requests will be blocked if all your credits have been consumed.' / 'You'll keep all the credits you've already paid for when switching tiers … Your rollover cap resets to 2x your new monthly plan rate.' / call duration '$0.06 per minute' with telephony '$0.014 per minute when using a Cartesia-provided phone number'. No region is named on the page.Source: Cartesia - pricing

Official sources

The vendor pages we read for the claims on this page, what each one states, and the day we read it.

Cartesia - pricing ↗
Cartesia pricing lists subscriptions and monthly credits; the page does not provide a country-by-country declared availability matrix. Historical registered scope read on 2026-10-15, not rechecked in full on 2026-10-15: Historical registered scope from 2026-10-12; the entire statement was not re-read on 2026-10-15: Free '$0 /mo' with '20K credits / month'; Pro '$5 /mo' with '100K credits / month', 'Commercial use license' and 'Instant voice cloning'; Startup '$49 /mo' with '1.25M credits / month' and 'Pro voice cloning'; Scale '$299 /mo' with '8M credits / month'; Enterprise 'Custom'. Sonic-3.6 'Minutes included monthly ~27 ~133 ~1,667 ~10,667' across Free, Pro, Startup and Scale, with 'Concurrent requests 2 3 5 15'. FAQ: 'Sonic text-to-speech: One minute of audio generation requires 750-800 credits. 1 credit equals 1 character.' / 'Your subscription will renew monthly from your purchase date.' / 'At pro tier, it's $65 per 1M credits.' / 'If your overages are off, new requests will be blocked if all your credits have been consumed.' / 'You'll keep all the credits you've already paid for when switching tiers … Your rollover cap resets to 2x your new monthly plan rate.' / call duration '$0.06 per minute' with telephony '$0.014 per minute when using a Cartesia-provided phone number'. No region is named on the page. Official body read on 2026-10-15, limited to this scope: Pro lists $5/month, 100K credits and about133 Sonic-3.6 minutes, commercial licence andTTS concurrency3. With overages off, requests stop when credits run out. Voice agents are separately metered at $0.06/call minute; telephony can cost extra. No region is named. Pro is listed at $5 per month with 100K credits; the page names no purchase region. This is a published plan quote, not a tested account checkout. Historical scope read on 2026-10-12, not rechecked in full on 2026-10-15: Free '$0 /mo' with '20K credits / month'; Pro '$5 /mo' with '100K credits / month', 'Commercial use license' and 'Instant voice cloning'; Startup '$49 /mo' with '1.25M credits / month' and 'Pro voice cloning'; Scale '$299 /mo' with '8M credits / month'; Enterprise 'Custom'. Pro lists 100K credits, about 133 Sonic-3.6 minutes and three concurrent TTS requests. With overages disabled, new requests stop when the balance is exhausted; voice-agent call minutes and telephony are separately metered. Historical scope read on 2026-10-12, not rechecked in full on 2026-10-15: Free '$0 /mo' with '20K credits / month'; Pro '$5 /mo' with '100K credits / month', 'Commercial use license' and 'Instant voice cloning'; Startup '$49 /mo' with '1.25M credits / month' and 'Pro voice cloning'; Scale '$299 /mo' with '8M credits / month'; Enterprise 'Custom'. Sonic-3.6 'Minutes included monthly ~27 ~133 ~1,667 ~10,667' across Free, Pro, Startup and Scale, with 'Concurrent requests 2 3 5 15'. FAQ: 'Sonic text-to-speech: One minute of audio generation requires 750-800 credits. 1 credit equals 1 character.' / 'Your subscription will renew monthly from your purchase date.' / 'At pro tier, it's $65 per 1M credits.' / 'If your overages are off, new requests will be blocked if all your credits have been consumed.' / 'You'll keep all the credits you've already paid for when switching tiers … Your rollover cap resets to 2x your new monthly plan rate.' / call duration '$0.06 per minute' with telephony '$0.014 per minute when using a Cartesia-provided phone number'. No region is named on the page.
Read on October 15, 2026
Cartesia - Terms of Service ↗
'Last Revised on June 14, 2024', for the website 'operated on behalf of Cartesia AI, Inc.' Section 4.1: 'We hereby permit you to use the Services for your personal, non-commercial use only (unless commercial use is expressly permitted by your subscription tier)'. Section 5.3(b): 'You represent and warrant that you will not monetize, make commercial use of, or otherwise use for or in connection with any commercial purposes, any of Your Outputs that you create with the Services (unless commercial use is expressly permitted by your subscription tier).' Section 5.3(c) 'Models': 'Inputs, Outputs, and user interactions with the Services may be used by Cartesia to train, enhance, evolve and improve its machine learning and artificial intelligence models and algorithms (collectively, "Models"), including for labeling, classification, content moderation and model training purposes.' Section 5.3(d) offers a form to opt out of training on selected categories of content, without undoing earlier use. Section 5.3: 'Cartesia does not claim ownership of any of your Inputs to the Services ("Your Inputs") or any of the Outputs you create with the Services ("Your Outputs").' Section 2.2: 'you agree to pay us the applicable fees and taxes in U.S. Dollars'. Section 2.4: cancel 'by clicking the cancel subscription button under the subscription tab in the playground at play.cartesia.ai/subscription'. Section 1: users must be 18 or older. Section 5.3(b) adds that Outputs may be downloaded from some but not all of the Services, where 'you are permitted to use such Output outside of the Services but always subject to these Terms'.
Read on October 12, 2026
Cartesia - API reference, text to speech bytes ↗
The request body requires 'output_format', documented as 'WAVOutputFormat · object' with the alternatives 'WAVOutputFormat MP3OutputFormat RAWOutputFormat', and the examples send '"output_format": { "container": "wav", "encoding": "pcm_s16le", "sample_rate": 44100 }' to 'https://api.cartesia.ai/tts/bytes'. The response is 'audio/*'. The page documents the API rather than an export from the playground.
Read on October 12, 2026
Cartesia - realtime text-to-speech quickstart ↗
'Stream text to Cartesia over a WebSocket and receive audio chunks back in real time.' An API key is created at play.cartesia.ai/keys, and the page warns that browser use should authenticate with ephemeral access tokens.
Read on October 12, 2026
cartesia official car-api ↗
Cartesia subscriptions include monthly model credits. TTS bytes/SSE/WebSocket requests consume about one credit per character; successful requests consume credits and errors do not. Managed Agents are billed separately per minute in USD without using the model-credit balance. Overage access depends on the workspace setting.
Read on October 15, 2026

Compare with other tools in Audio Generation

Product	Maker	Pricing	Launched	Shared tags
Typecast	Neosapience	Freemium	Launch date not verified	Voice, Text-to-speech, Commercial use
ElevenLabs	ElevenLabs	Freemium	2023-01	Voice, Text-to-speech
Amazon Polly	Amazon Web Services	Paid	Launch date not verified	Voice, Text-to-speech
Udio	Udio	Paid	2024-04-10	—
Suno	Suno	Freemium	2023-12-20	—
Stable Audio	Stability AI	Freemium	2023	—