CartesiaUnclaimed

Cartesia builds real-time voice AI. It offers speech models for text-to-speech and speech-to-text, and Managed Agents, a platform for voice agents that answer and make phone calls. The models and agents are available through one API and are built to be fast enough for live conversation.

The main products are Sonic, a text-to-speech model, and Ink, a speech-to-text model. Managed Agents runs AI voice agents for phone calls and uses Cartesia's own models. The company says its team pioneered state space model architectures, including Mamba, and it runs its models on state space models for low latency at scale.

The company says its models can be deployed in the cloud through regional API endpoints, on-premise, or on-device. Its site says inference can run in-region to help with latency and data residency requirements.

The site describes Cartesia's speech models as supporting 44 languages. It also says EliseAI runs more than 5 million voice calls a month on Cartesia.

Pricing starts with a free plan that includes 20K credits a month and $1 of prepaid agent usage. The Pro plan costs $5 a month, Startup costs $49 a month, and Scale costs $299 a month. Enterprise pricing is custom and includes items such as SSO, DPAs and BAAs. Managed Agents is listed at $0.03 per minute, with telephony at $0.014 per minute when using a Cartesia-provided phone number. LLM usage during calls is billed separately.

The page says Professional Voice Clone training is free. Credits are used for speech generation, and paid plans can turn on overages at a set rate per million credits.

The company offers an API, documentation, cookbooks and an MCP server for developers. Teams that want to talk to sales can contact the company directly, and the free plan lets people start building without a sales call.