The upgraded T1 mannequin affords sooner response instances and enhanced capabilities for processing prolonged textual content paperwork, the corporate mentioned in a publish on its official WeChat account.
T1 can “hold the content material logic clear and the textual content neat and clear”, the publish mentioned, whereas the hallucination charge is “extraordinarily low”.
The launch comes amid heightened rivalry in China’s AI panorama following DeepSeek’sDeepSeek’s introduction of fashions that provide comparable or superior efficiency to Western techniques at considerably decrease prices.
Tencent had beforehand launched a preview model of T1 by platforms, together with its AI assistant utility Yuanbao.
The official model shall be powered by Tencent’s Turbo S foundational language mannequin, unveiled late final month, which the corporate claims processes queries sooner than rival DeepSeek’s R1 mannequin.
A chart revealed within the publish evaluating the T1 mannequin to DeepSeek R1 confirmed Tencent’s outperformed on some data and reasoning benchmarks.
Tencent has ramped up its AI investments in current months. On Thursday, the corporate introduced plans to extend capital expenditure in 2025, following already aggressive AI spending all through 2024.
© Thomson Reuters 2025
