NewsBite analysis · AI, Cloud, Tools
MarkTechPost
A Streaming TTS Model That Now Leads Both Artificial Analysis Speech Arenas
Cartesia has introduced Sonic-3.6, a new version of its real-time text-to-speech model, which has claimed the top spot on two major Artificial Analysis speech leaderboards. This update follows Sonic-3.5 and significantly boosts the model's naturalness, with an incredible latency of under 90 milliseconds from text to audio output. Built on state space models rather than transformers, Sonic-3.6 is available exclusively as a beta hosted API, offering features like instant voice cloning and customizable pronunciation dictionaries.
Act nowConfidence: HighShort term horizon
Published 18 Aug 2026, 10:41 UTC2 min read
Visit original article ↗Lead image — MarkTechPost
Summary
TL;DR
- Sonic-3.6 ranks #1 on both Provider Voice and Controlled Voice leaderboards.
- The model demonstrates significant improvements in naturalness and speed.
- It is available as a closed commercial model, with no open weights.
- Features include inline expression tags and customizable voice parameters.
- Debuts with pricing competitive to existing models in the market.
Interesting facts
- The model can achieve sub-90ms latency, a significant technical achievement.
- Sonic-3.6 utilizes a state space model architecture, distinct from transformer-based systems.