Tencent Launches Hunyuan 2.0, Its Next-Generation AI Model
Tencent has officially released Hunyuan 2.0, its latest self-developed large artificial intelligence model, marking a significant advancement in its AI capabilities. The new model is available in two distinct versions: Think and Instruct. Built on a mixture-of-experts (MoE) architecture, Hunyuan 2.0 boasts 406 billion total parameters with 32 billion active parameters, and supports an ultra-long context window of 256K tokens. Compared to its predecessor, the Hunyuan 2.0 Think version features substantial improvements in pretraining data and reinforcement learning strategies, resulting in enhanced performance in complex reasoning tasks across mathematics, science, and coding domains. The model has already been integrated into Tencent’s AI-native products, such as Yuanbao and ima, and is accessible to developers and enterprises via Tencent Cloud’s APIs. This release allows users to leverage advanced reasoning and content creation capabilities. The announcement highlights Tencent's continued investment in proprietary AI infrastructure to compete in the rapidly evolving global technology landscape, providing robust tools for both internal applications and external commercial use through cloud services.
Wire timeline
Tencent Launches Hunyuan 2.0, Its Next-Generation AI Model
Tencent has officially released Hunyuan 2.0, its latest self-developed large artificial intelligence model, marking a significant advancement in its AI capabilities. The new model is available in two distinct versions: Think and Instruct. Built on a mixture-of-experts (MoE) architecture, Hunyuan 2.0 boasts 406 billion total parameters with 32 billion active parameters, and supports an ultra-long context window of 256K tokens. Compared to its predecessor, the Hunyuan 2.0 Think version features substantial improvements in pretraining data and reinforcement learning strategies, resulting in enhanced performance in complex reasoning tasks across mathematics, science, and coding domains. The model has already been integrated into Tencent’s AI-native products, such as Yuanbao and ima, and is accessible to developers and enterprises via Tencent Cloud’s APIs. This release allows users to leverage advanced reasoning and content creation capabilities. The announcement highlights Tencent's continued investment in proprietary AI infrastructure to compete in the rapidly evolving global technology landscape, providing robust tools for both internal applications and external commercial use through cloud services.
TechNode