Sonic 3.6
Low-latency streaming TTS for voice agents.
Sonic 3.6 is Cartesia’s streaming text-to-speech model, built for real-time voice agents on Cartesia’s state-space-model research.
- Provider
- Cartesia
- Type
- Text-to-speech
- Released
- Aug 17, 2026
- Price
- ≈$49 per 1M characters
- Input
- text
- Output
- audio
- Open weights
- No
Best for
- Real-time / low latency
- Agents
- Multimodal
Strengths
- Built for real-time agents
- Top-5 speech arena Elo
Tools that use it
- Cartesia — Cartesia. Ultra-low-latency voice API for real-time agents.