Cartesia's monthly price buys two budgets: credits for Sonic text to speech and Ink speech to text, and the same amount again in prepaid dollars for Managed Agents ($5 on Pro, $49 on Startup, $299 on Scale). Estimate model usage in credits and calls in minutes; for a live voice agent, the number of simultaneous streams often decides the plan before either budget runs out.
Cartesia's own conversions put a minute of text to speech at about 750 credits on every plan (Pro's 100,000 credits make about 133 minutes) and a minute of speech to text at about 180 credits (Pro's credits make about 9 hours 16 minutes). Both come out of one pool, so Pro covers 133 minutes of speech or about 9 hours of transcription, not both. In ToolColumn's arithmetic, a minute of speech costs about 3.8 cents on Pro, 2.9 cents on Startup and 2.8 cents on Scale.
Startup's 1.25 million credits cover about 1,667 minutes of speech a month, but it allows 5 simultaneous text-to-speech requests against Scale's 15. An agent that may speak on six calls at the same moment needs Scale even when Startup's credits would last the month, so for that agent the step from $49 to $299 pays for ten more streams, not the extra credits. Startup fits a product that stays within five simultaneous streams and needs more than Pro's 133 minutes of speech.
Managed Agents costs $0.03 per minute of call duration on every plan, down from the $0.06 Cartesia previously listed, plus $0.014 a minute on a Cartesia phone number; LLM usage during calls is billed separately. In an illustrative month, an agent taking 1,000 minutes of calls on a Cartesia number runs up $44 in agent and telephony charges before LLM usage. In ToolColumn's arithmetic, Startup's $49 of prepaid agent dollars equals about 1,110 such minutes and Scale's $299 about 6,800.