PolyAI Launches Dialog-RSN-1 Voice Model, Beats GPT and Gemini in Enterprise Tests

matthen2 · x · 2026-07-31

PolyAI released Dialog-RSN-1, an end-to-end audio LLM capable of handling turn-taking, reasoning, response, and transcription simultaneously.

Trained on millions of real enterprise conversations, the model reportedly outperformed all real-time-capable OpenAI and Google models on audio understanding and safety in a benchmark for real customer calls, while responding in under 300 milliseconds.

Related event: PolyAI Launches End-to-End Native Audio Model Dialog-RSN-1(4 posts)→

Original post →

More from Models

Models channel →