Tencent Unveils Advanced T1 Reasoning Model, Escalating the AI Arms Race in China

The Chinese technology behemoth Tencent unveiled the official release of its T1 reasoning model on Friday evening, intensifying the competition in China’s bustling artificial intelligence arena.

The enhanced T1 model boasts quicker response capabilities and improved functionality for handling extensive text documents, as mentioned in a message on the company’s official WeChat channel.

According to the announcement, T1 is capable of «maintaining clear logical content while keeping the text organized and tidy», with a «very low» occurrence of inaccuracies.

This rollout comes in the midst of intensified competition within the Chinese AI market, following DeepSeek’s launch of models that provide performance on par with or surpassing Western counterparts at significantly reduced prices.

Before this release, Tencent had already shared a beta version of T1 across various platforms, including its AI assistant app Yuanbao.

The official version will utilize Tencent’s Turbo S foundational language model, introduced just a month ago, which the company asserts processes inquiries more swiftly than competing models like DeepSeek’s R1.

A comparison chart included in the announcement indicated that Tencent’s model excelled in several knowledge and reasoning evaluations compared to the DeepSeek R1.

In recent months, Tencent has significantly increased its AI investments. Just yesterday, the company revealed intentions to boost capital expenditure in 2025, following an already robust AI spending strategy throughout 2024.

© Thomson Reuters 2025