Tencent Hunyuan announced today that its all-new real-time voice model, Hy Realtime, has been officially released. As a voice model built around real-time interaction, it performs strongly in core conversational abilities such as speech understanding, instruction following, and human-like expression, and it has also added two new skills in one go: first, agentic tool calling; second, singing. In other words, it is no longer just a "mouth" that simply responds, but has begun to grow hands capable of doing work, bringing it one step closer to "communicating like a human."
Hy Realtime is already quietly running in products such as Yuanbao voice calls and Tencent customer service, and it will soon be made available on Tencent Cloud as an API open to external users. Internally, it directly improves the conversational experience for hundreds of millions of users; externally, it marks a key step for Tencent in bringing its self-developed voice capabilities from internal scenarios to external developers. Similar capabilities were previously mostly provided by third parties.
