Cartesia Sonic 3

Cartesia

Cartesia's real-time text-to-speech model (late 2025), built on a state-space architecture for extremely low latency (~90ms model latency) with strong emotional expressiveness, multilingual support, and instant voice cloning from short samples - ideal for live voice agents.

Strengths

Ultra-low latency (~90ms model, ~190ms end-to-end)
State-space architecture for efficient real-time synthesis
Strong emotional expressiveness across 40+ languages
Instant voice cloning from short samples
Purpose-built for live voice agents

Caveats

Proprietary - API access only
Usage-based pricing scales with volume
Voice cloning gated by consent policies

Vendor: Cartesia
License: Proprietary
Modalities: text
audio

Capabilities

Vision

Audio

Video

Tool Use

Resources

API Documentation

Pricing

Usage-based API pricing by characters/credits with subscription tiers; optimized for real-time voice agents.