Voice & audio

Cartesia

Cartesia builds speech models and the voice agents that run on them, aimed at teams putting live conversation into products and contact centers.

Best for Developers building low-latency voice agents and real-time speech features

Starts at
$5/month flat
Free plan
Yes
Price checked
2026-10-19

Pricing

Price checked: 2026-10-19. Source: official pricing page.

Free plan: 20K credits/month and $1 prepaid agent balance; commercial-use license starts on Pro

PlanMonthlyAnnual (per month)Billing unitNotesSource
Pro$5Not listedflat100K credits/mo; commercial license; instant voice cloningSource
Startup$49Not listedflat1.25M credits/mo; pro voice cloningSource
Scale$299Not listedflat8M credits/moSource
EnterpriseNot listedNot listedflatCustom; contact salesSource

"Not listed" means the official page shows no list price for that billing period (often contact sales). We never fill gaps from third-party sites.

Overview

Developers call Sonic for text-to-speech, Ink for speech-to-text and Managed Agents for phone bots that can take or make calls, with provisioned numbers and concurrency limits that grow by plan. The company leans on state space model research to keep latency low enough for natural back-and-forth, and markets coverage of 44 languages. Plans bundle a monthly credit pool for the speech models plus a prepaid balance for agents, so one subscription covers both. Voice cloning comes in instant and professional forms on higher tiers. Limitation: commercial use is excluded until you pay, and agent minutes and telephony are billed on top, so call-heavy deployments need careful forecasting. Regulated industries should confirm DPA or BAA terms, which sit on Enterprise.

Key features

Limitations

The free plan has no commercial-use license. Agent calls and phone time are billed per minute on top of plan credits.

Links