Cartesia Pricing Explained: Credits, Agent Minutes and Extras

Cartesia plans run from free to $299 a month. Sonic uses about 1 credit a character and Ink 2 uses 3 a second. Managed Agents is $0.03 a minute, plus phone and model. Dograh, the open-source AI voice agent platform we build, runs Sonic and Ink on your plan, and you can self-host it or run it on Dograh Cloud.

Key Takeaways

• Cartesia bills voices by the character and speech-to-text by the second, silence included.
• Managed Agents is $0.03 a minute, before the phone number and language model.
• Each plan caps concurrent agent calls, from 8 on Free to 60 on Scale.

A Cartesia bill mixes speech credits with dollars for agent minutes, and the language model sits outside both. We walk through each plan and meter, price a Managed Agents minute, run Cartesia's voices in Dograh on your plan and show where the bill grows. For the wider field, see our guide to Cartesia alternatives.

Cartesia's plans at a glance

Every Cartesia plan renews monthly and comes with two balances, credits for Sonic and Ink and prepaid dollars for agent minutes. On paid plans the agent dollars match the plan price. The figures below come from Cartesia's pricing page.

Commercial use and Instant voice cloning start on Pro. Professional voice clones need Startup, with two slots, or Scale, with four. Enterprise adds custom volume pricing, SSO, data processing agreements and business associate agreements (BAAs).

How credits pay for Sonic and Ink

Sonic uses about 1 credit for each character it speaks, and speech from a cloned voice costs the same. Cartesia puts a minute of audio at roughly 750 to 800 credits.

Speech-to-text is billed by the second of audio, and silence counts. According to Cartesia's credit rates, Ink 2 streaming uses 3 credits a second and ink-whisper uses 1. An hour of Ink 2 is 10,800 credits, which Cartesia puts at $0.39 on Scale, and an hour of ink-whisper is 3,600 credits.

Our own arithmetic from the plan prices puts 1,000 characters of Sonic at about $0.050 on Pro, $0.039 on Startup and $0.037 on Scale. When credits run out, requests fail unless you have turned overages on. Overages then cost $65 per million credits on Pro, $45 on Startup and $38 on Scale.

What a Managed Agents minute costs

Cartesia lists Managed Agents at $0.03 a minute, down from $0.06. That rate is billed in dollars, outside your credits, and paid first from the plan's prepaid agent dollars.

The $0.03 leaves two things out. A Cartesia phone number adds $0.014 a minute, so a call on one of its numbers starts at $0.044 before the model. The language model is billed separately at the price Cartesia lists for each model in its catalog. Numbers on your own Twilio account or SIP trunk skip the $0.014, and your carrier bills those minutes instead.

The prepaid agent dollars cover about 33 minutes on Free, 2 hours 47 minutes on Pro, 27 hours on Startup and 166 hours on Scale, before the phone line and the model. After that, agent minutes draw on your billing method.

Open Source Alternative to Vapi / Retell

Self-hosted voice agent platform — no per-minute fees

dograh-hq/dograh

Star on GitHub

Cartesia voices inside Dograh

You can also keep Cartesia's speech and change the platform. Dograh has Cartesia as a built-in provider in its voice settings, so Sonic and Ink run on your own Cartesia plan at its credit rates, and the agent minute is no longer billed by Cartesia.

Our arithmetic for one call minute uses 400 characters of agent speech, the same assumption as our other pricing posts, at Scale's credit rate. Self-hosted Dograh has no per-minute platform fee, but servers, model keys and phone minutes still cost money. Dograh Cloud charges a 1¢-a-minute platform fee.

Cartesia's concurrency limits still apply when Dograh calls it, so check that your plan allows as many voice requests as you run calls at once. Dograh can hold several Cartesia keys and spread requests across them, which our post on speech-to-text and voice concurrency limits explains.

Where the bill grows

The language model is the easiest line to miss, because it is billed outside both the credits and the $0.03. Phone minutes come next, from Cartesia or your own carrier. Overages and Professional clone slots add to it as well.

The voice meter shrinks when the agent speaks less from scratch. Recording fixed lines such as greetings and disclosures, and generating only the rest, cuts the characters you pay for, as our post on mixing pre-recorded audio with TTS shows. Our AI voice agent cost per minute breakdown covers the same layers across other vendors.

Which plan fits a phone agent

Free and Pro suit testing, with 2 or 3 voice requests at a time. Startup is the first plan with Professional voice clones, and Scale is the one built for production volume, with 60 agent calls and 15 voice requests at once. Enterprise is where custom concurrency, regional deployments and compliance terms come in.

Before you pick, write down your real characters per minute, call length and calls at once, then price both setups. Our Cartesia vs ElevenLabs comparison sets these prices next to ElevenLabs' if you are weighing both vendors.