Disclosure: some links on this page may be affiliate links; we may earn a commission at no extra cost to you. Read the disclosure.

By the toolist.ai editors. Researched from Cartesia's own pages and hands-on tests published by others, not hands-on tested by toolist.ai. Updated November 2, 2026.

Cartesia review: pricing, features and what testers found

Cartesia makes real-time text-to-speech and speech-to-text models, voice cloning and voice agents, sold through a web playground and a developer API. A free plan gives 20,000 credits a month (about 27 minutes of speech); paid plans start at $5 a month.

Our take

Researched from public sources, not hands-on tested by toolist.ai. Cartesia suits developers building live voice agents more than creators making narration: its Sonic-3.6 model streams speech through an API and a web playground, Pro costs $5 a month for about 133 minutes of speech, and on Artificial Analysis's blind-vote leaderboard on November 2, 2026 Sonic-3.6 ranked 4th of 100 models. SaaS24reviews' Supriya Siddhanta ran the same script through Cartesia's older Sonic-3 and ElevenLabs in January 2026 and found ElevenLabs more natural for long-form narration, with less fine control over pacing and emphasis in Cartesia. The catches: Cartesia's own latency claim (under 90 ms) is far below the 437 ms median that Coval's independent benchmark measured for Sonic-3.6 in September 2026, and Cartesia's pages disagree on whether the free plan includes instant voice cloning.

Based on Cartesia's published plans and the hands-on tests linked below, not on a test of our own.

Good for

  • Developers building voice agents, support bots or phone assistants where speech has to start quickly
  • Trying a modern text-to-speech model cheaply: Pro is $5 a month for about 133 minutes of speech with a commercial licence
  • Multilingual agents, including 11 Indic locales and Hindi-English code-switching in one generation

Watch out for

  • Latency claims differ by who measures: Cartesia says Sonic-3.6 replies in under 90 ms, but Coval's independent benchmark put its median perceived time to first audio at 437 ms in early September 2026, slower than Cartesia's own Sonic-3.5 (270 ms) and ElevenLabs Flash v2.5 (182 ms) on the same board.
  • Cartesia's pages disagree on the free plan: the Sonic page says Free includes instant voice cloning, but the pricing page lists instant cloning from Pro and marks it unavailable on Free. Free is for evaluation and personal use; the commercial licence starts at Pro.
  • Minutes are estimates: credits are spent at 1 per character, which Cartesia puts at about 750 to 800 credits a minute of audio, so the plan minutes this record uses (Cartesia's own approximations) are the upper end.
  • Past the monthly credits, paid plans either stop or, with overages on, charge $65 (Pro), $45 (Startup) or $38 (Scale) per million credits; unused credits roll over up to twice the plan's monthly credits.
  • SaaS24reviews' tester, trying the older Sonic-3 in January 2026, found its delivery slightly less refined than ElevenLabs for polished narration, so creators of audiobooks or voiceovers may prefer a narration-focused tool.

What Cartesia does

Price

Free plan: 20,000 credits a month (about 27 minutes of Sonic-3.6 speech) and $1 of voice-agent usage; for evaluation and personal use only, no commercial licence

PlanBilled monthlyBilled yearly, per monthIncludesPrice per minute of generated speech
Free free not listed 20,000 credits a month (about 27 minutes of Sonic-3.6 speech, about 1 hour 51 minutes of speech to text) and $1 of voice-agent usage; 2 concurrent TTS requests, 1 phone number; evaluation and personal use only –
Pro $5 not listed 100,000 credits a month (about 133 minutes of speech) and $5 of voice-agent usage; commercial use licence, instant voice cloning, 3 concurrent TTS requests, 3 phone numbers; overages $65 per million credits $0.038
Startup $49 not listed 1.25 million credits a month (about 1,667 minutes of speech) and $49 of voice-agent usage; 2 professional voice clone slots, organizations, 5 concurrent TTS requests; overages $45 per million credits $0.029
Scale $299 not listed 8 million credits a month (about 10,667 minutes of speech) and $299 of voice-agent usage; 4 professional voice clone slots, priority support, 15 concurrent TTS requests; overages $38 per million credits $0.028
Enterprise not listed not listed Custom credits, agent usage and volume pricing; custom concurrency, DPAs and BAAs, a shared Slack channel, SSO and security questionnaires –

Price per minute of generated speech is our calculation: the monthly-billing price divided by what the plan includes, assuming you use all of it. It is not a measured cost.

Prices last checked: 2026-11-02 on Cartesia's pricing page. Full breakdown: Cartesia pricing.

Price history

No price change since we first recorded these prices on November 2, 2026. We re-check them every week and list each change here.

What hands-on testers found

We have not tested Cartesia ourselves. These are the published tests we found that used it hands-on, newest first; follow the links for the full reviews.

Artificial Analysis Speech Arena · November 2, 2026

What they tested: Blind preference votes between text-to-speech models, read on the day we checked.

  • Sonic-3.6 ranked 4th of 100 models (Elo 1296), behind Eleven v4 Turbo, Eleven v4 and Qwen-Audio-3.1-TTS-Plus.
  • The older Sonic 3.5 ranked 21st and Sonic 3 64th.

Their score: Rank 4 of 100 models (Sonic-3.6)

SaaS24reviews · Supriya Siddhanta · updated April 24, 2026

What they tested: Cartesia's text to speech (the article describes Sonic-3) in its web dashboard, with the same script run through ElevenLabs, publishing both audio samples (a documentary script; the 71-second Cartesia clip reads its opening) and dashboard screenshots. The article was first published on January 27, 2026 and its audio uploaded on January 26 and 27, so the test used Sonic-3, before Sonic-3.5 and Sonic-3.6. It does not name the voice or plan, and its latency remarks are impressions, not measurements; its Cartesia prices ($4, $39 and $239 a month) are the yearly-billed rates Cartesia showed in January 2026 for the same plans and credits, and Cartesia now lists monthly prices only. The page has no links to Cartesia; the site's general disclosure says some links may be affiliate links.

  • On the same script, ElevenLabs sounded more natural and human than Cartesia for long-form content.
  • Cartesia's delivery suits conversational agents but felt slightly less polished than ElevenLabs for narration.
  • Cartesia gives less control over pacing and emphasis than ElevenLabs; she calls it closer to choosing a style and going.
  • She rated Cartesia the better pick for real-time voice agents and developers, and ElevenLabs for creators.

Test result

toolist.ai has not run its own test of Cartesia; the findings above come from the published tests linked. See how we research and test.

Best voice tools · Cartesia pricing

Do you make Cartesia? Claim this listing to correct it; it's free. To see whether ChatGPT, Claude, Gemini and Perplexity recommend it, and how many readers we send you: toolist.ai for AI companies.

Sources