Tencent Yuanbao has ushered in a major upgrade and has access to two major models at the same time: the official version of the deep thinking model "Tencent Hunyuan T1" and the latest version of DeepSeekV3-0324. After the upgrade, Yuanbao's abilities have been comprehensively enhanced, and it can perform even better in scenarios such as programming, mathematical reasoning, and long text processing.

Among them, the official version of Hunyuan T1 is upgraded based on the new generation of fast thinking model TurboS base. It is a strong reasoning model that can reply in seconds, speak quickly, and is good at processing ultra-long texts. It innovatively adopts the hybrid Mamba architecture and maintains industry-leading performance. It is suitable for processing various deep thinking tasks such as academic papers, planning plans, and long-form content;

The latest version of DeepSeekV3 deployed by Yuanbao is good at rapid generation. You can write code with one sentence of instructions. It has stronger front-end development capabilities and realizes website and app development. The performance of solving math problems and text creation is also faster and more accurate than the previous generation, and it is capable of various high-frequency usage scenarios.

It is worth mentioning that Tencent Yuanbao was one of the first AI applications to access the DeepSeekV3-0324 version, and it only took 1 day from model open source to online.

Currently, Tencent Yuanbao’s “Hunyuan + DeepSeek” dual models both support deep thinking and instant answers. Recently, Yuanbao has continued to update frequently, iterating 30 versions in 35 days, and continuously launching practical functions. All abilities are free and open for unlimited use, covering more work, study and life scenarios.