Tencent Hunyuan announced on August 31 that since the launch of its flagship model Hy4preview on the WorkBuddy platform on August 28, it has quickly sparked a wave of user and developer experiences due to significant improvements in its Agent capabilities. Due to the high concurrency demand on the day of release, the WorkBuddy task queue experienced waiting queues. To address this, Tencent has urgently expanded the Hy4preview inference cluster and will continue to dynamically allocate resources to ensure service response efficiency.
As large model technology accelerates its application in complex Agent scenarios, the supply and demand gap for high-performance inference computing power is becoming increasingly prominent. Tencent stated that due to the limited total scale of high-end computing power and the high concentration of concurrent calls during peak times, short-term waiting situations may still occur during specific peak periods. To alleviate traffic peaks and meet continuous business needs, the official also introduced a transitional alternative solution, extending the free usage rights of existing models such as Hy3 on WorkBuddy until 23:59 on September 30, 2026.
The sudden surge in call volume caused by the enhanced Agent capabilities of Hy4preview reflects that the practical potential of AI Agents in real productivity scenarios is being accelerated. As production-level Agents demand higher real-time reasoning and multi-step planning capabilities, the ability to flexibly expand computing clusters and coordinate multiple models is gradually becoming a core competitive advantage for AI infrastructure and platform service providers.
