Tencent Hunyuan Releases Real-Time Voice Model Hy Realtime

On November 13, 2026, Tencent Hunyuan unveiled Hy Realtime, a new real-time voice model. According to IT Home's report that afternoon, the model delivers strong performance in basic conversational tasks such as speech understanding, instruction following, and human-like expression, while adding new Agentic tool-calling and singing capabilities. The company said Hy Realtime can restore a user's intent from incomplete inputs such as slurred speech, half-finished sentences, or mid-sentence corrections by drawing on conversational context, significantly improving understanding accuracy. Its multi-turn instruction following and context-management abilities were also said to be strengthened.

A 36Kr report later the same day noted that Hy Realtime can decide whether to run a web search based on conversational context and user intent, unifying search decisions, tool calls, and response generation. The model is already live in voice calls on the Yuanbao app and in Tencent customer service, and will soon be available on Tencent Cloud through its API service. The current version can be tried in the latest Yuanbao app by selecting the Chuanyu voice tone in the voice-call feature, and a web-based Demo offers access to selected capabilities. Tencent Hunyuan added that Hy Realtime is still being iterated, with future work focused on full-duplex interaction, long-term personalized memory, and front-end and back-end Agent interaction, leaving the model at the stage of rolled-out deployment and ongoing refinement.