Cartesia

What it is for

Cartesia is a hosted realtime speech stack: type text into Sonic, get spoken audio, or run Ink transcription and managed voice agents on the same API. Use it when the job is voice that has to answer live, not when you need a library of celebrity clones.

It is not open source. Free exists ($0, 20K credits / month). Paid cards start at Pro $5/mo.

What makes it unique: the cover is Cartesia’s signed-out Sonic composer — a support script, Jacqueline / American English, Play — first audio is pitched under 90 ms.

Key features

• Sonic text-to-speech — type a script, pick a voice and language, press Play; Cartesia says first audio lands in under 90 ms (cartesia.ai; Sonic).
• Ink speech-to-text — streaming transcription on the same credit pool as TTS (Pricing).
• Instant voice cloning — Pro unlocks a clone you can speak with immediately (Pricing).
• Professional Voice Clone — Startup+ trains a PVC; generated speech is still billed per character (Pricing FAQ).
• Managed voice agents — build agents that answer and place calls on Cartesia-provisioned numbers (cartesia.ai; Pricing).
• Cloud, on-prem, or on-device — the same models can run in-region or on hardware you own (cartesia.ai).

Who it's for

Best fit: product teams shipping a live voice agent or TTS API who will stay on Sonic / Ink.

Poor fit: narrators who only need a one-off MP3 of a script (WellSaid).

Pricing

Verified 2026-10-22 (Asia/Shanghai) from cartesia.ai/pricing. Credits roll up to 2× the monthly rate. Re-check the calculator before you buy.

• Free — $0. 20K credits / month. $1 prepaid agents / month. TTS + STT. No commercial-use license on this card.
• Pro — $5/mo. 100K credits. $5 prepaid agents. Commercial use + instant voice cloning.
• Startup — $49/mo. 1.25M credits. $49 prepaid agents. Pro voice cloning + organizations. 2 PVC slots.
• Scale — $299/mo. 8M credits. $299 prepaid agents. Priority support + higher concurrency. 4 PVC slots.
• Enterprise — custom credits, SSO, DPA/BAA, shared Slack.

1 credit ≈ 1 TTS character. Ink STT is 3 credits per second of audio. Agent minutes are $0.06/min against the prepaid-agent dollars; Cartesia phone numbers add $0.014/min. Overages (if enabled): Pro $65 / 1M credits; Startup $45; Scale $38.

Limitations / costs

• Free is not a commercial license — Pro is the first card that says so.
• Credits and agent dollars are two meters. A Scale $299 plan is 8M model credits and $299 of agent prepaid, not unlimited calls.
• PVC slots are plan-gated (Startup 2 / Scale 4). Free and Pro do not include a Professional Voice Clone.
• We did not embed a third-party or deprecated-channel video. Official channel is linked below.

Alternatives

• ElevenLabs — broader voice marketplace if you need a huge clone catalog, not a 90 ms Sonic stack.
• Murf AI — studio TTS for explainer videos if you are not shipping a phone agent.
• WellSaid — narrator voices for training video, not live agents.

Official tutorial

We searched YouTube for a current official Cartesia getting-started video. @CartesiaAI / @cartesiaai is the live channel (the older @cartesiatv handle is stale). We did not embed a video from a deprecated or third-party channel. Use the official channel and the written Pricing plus cartesia.ai.

Website

https://cartesia.ai

Source

Catalog row curated for AI Tools Nav. Cover is Cartesia’s signed-out Sonic Play composer. Free $0 / 20K credits. Pro $5; Startup $49; Scale $299. Official channel: @CartesiaAI (no stale-channel embed).

Last verified

2026-10-22 (Asia/Shanghai) editorial check.