By Yelin Mo and Andrew Silver
SHANGHAI (Reuters) – The Chinese Technology giant Tenncent launched the official version of its T1 reasoning model on Friday evening, intensifying competition in the artificial intelligence sector increasingly congested in China.
The upgraded T1 model offers faster response times and improved capacities for the processing of extended text documents, said the company in an article on its official WeChat account.
T1 can “keep the logic of clear content and clean and clean text”, according to the post, while the hallucination rate is “extremely low”.
The launch comes in the middle of increased rivalry in the Chinese AI landscape after the introduction of Deepseek models which offer performance comparable or superior to Western systems with significantly lower costs.
Tencent had previously published a T1 preview version via platforms, including its AI Yuanbao assistant application.
The official version will be fed by the fundamental language model of Tencent Turbo, unveiled at the end of last month, which, according to society, claims that the processes are faster than the R1 model of Deepseek Rival.
A graphic published in the post comparing the T1 model to Deepseek R1 has shown that Tencent has surpassed certain knowledge and reasoning references.
Tencent has increased its investments in AI in recent months. Thursday, the company announced its intention to increase capital spending in 2025, following the expenses of AI already aggressive throughout 2024.
(Report by Yelin Mo and Andrew Silver; edition by Shri Navaratnam)
