Rime voice models now available on Together AI

Together AI · 6 months ago

Rime text-to-speech models Arcana v2 and Mist v2 are now available on Together AI's platform, offering enterprise-grade voice synthesis with deterministic pronunciation control and production-grade latency. Mist v2 achieves approximately 225ms time-to-first-audio latency on dedicated endpoints and serves tens of millions of production calls monthly, while Arcana v2 provides 40+ voices across multiple languages trained on real customer service conversations. The integration allows voice agents to run STT, LLM, and TTS workloads on a single unified platform with consistent observability, reducing multi-vendor complexity and improving end-to-end conversational latency.