Tencent, the Chinese technology conglomerate, has officially launched its Hunyuan T1 reasoning model on March 21.

According to a statement released on the company's official Weixin (also known as WeChat) account, the upgraded Hunyuan T1 model is designed to deliver faster response times and enhanced capabilities in processing extensive text-based data.

Tencent claims the Hunyuan T1 model offers improved clarity and structural coherence in generated content while maintaining a notably low hallucination rate—an issue that has plagued many AI systems. 

The launch follows the recent introduction of high-performing models by domestic rival DeepSeek, which has positioned itself as a formidable competitor to Western AI systems at significantly reduced costs.

The official release of Hunyuan T1 comes after its preview version which was available in Yuanbao chatbot. According to Tencent, the reasoning ability of the official version is better than the preview version.

Image: Tencent/Weixin.

Tencent provided a comparative performance analysis, indicating that T1 outperforms DeepSeek R1 on select knowledge and reasoning benchmarks. 

The official version of this model is powered by Tencent's Turbo S foundational language model, introduced late last month, which the company claims delivers superior processing speeds compared to DeepSeek's R1 model.

The official version of Hunyuan T1 is based on the Hunyuan Turbo S, introduced late last month. Tencent claims that it is 'good at processing super long texts'.