Tencent reveals advanced AI developments to match competitors' technological advancements
Tencent, one of China's leading tech companies, has been making significant strides in the realm of artificial intelligence (AI). Over the past few months, the company's AI assistant feature in Tencent Meeting has seen a 20-fold increase in daily usage.
In addition to this, Tencent has been leveraging its AI capabilities across various industries, including finance, healthcare, education, automotive, and energy. One of their key offerings is Yuanqi, a product similar to GPTs that allows enterprises and developers to create digital agents using official Tencent plugins and knowledge bases.
Tencent Cloud has also released three new engines at the platform-as-a-service (PaaS) level, including the large model knowledge engine, the image creation engine, and the video creation engine. These engines are designed to empower developers to create more feature-rich applications.
One of Tencent's most notable AI models is Hunyuan. Since its release last year, Hunyuan has undergone steady updates, with the current model adopting a mixture of experts (MoE) structure, resulting in a 50% improvement in overall performance compared to the previous generation.
Hunyuan offers three model sizes: Pro, Standard, and Lite, available for both enterprises and individual developers. The open-sourced Hunyuan text-to-image model is also the first native Chinese open-source text-to-image model, adopting the same diffusion transformer (DiT) architecture as OpenAI's Sora.
Based on Hunyuan's video generation capabilities, users can input videos to immediately generate videos in specific styles with smooth and natural video frames and strong temporal consistency. This feature has the potential to revolutionize video production processes.
In the text-to-speech domain, Hunyuan's accuracy rate for the "needle in a haystack" test can reach up to 99.9%, demonstrating its exceptional capabilities in long text input scenarios.
Tencent's AI advancements extend to their messaging app, WeChat. A new feature in WeChat Read, based on the Hunyuan large model, enables users to ask AI about the themes of books without needing to read them entirely.
Perhaps one of the most anticipated developments from Tencent is the launch of a new assistant app called Yuanbao, expected by the end of May. Yuanbao is expected to address the gap in retail user access to Tencent's Hunyuan, offering a unified mobile entry point.
Lastly, Tencent's AI code assistant has been adopted by more than 50% of the development staff within the group, with a code generation rate exceeding 30% and a more than 20% improvement in R&D efficiency. This underscores Tencent's commitment to leveraging AI to streamline and optimize its development processes.
In comparison to ByteDance, which is more B2C-oriented, Tencent appears to be taking a more industry-oriented route with its large models. This strategic decision could potentially position Tencent as a major player in the AI industry.
While the exact publication date for the availability of Tencent's Yuanbao app is yet to be announced, it is clear that Tencent's AI advancements are reshaping the tech landscape, offering exciting possibilities for businesses and consumers alike.
Read also:
- Ford Discontinues Popular Top-Seller in Staggering Shift, Labeled as a "Model T Event"
- 2025 Witnesses a 27% Surge in Worldwide Electric Vehicle Sales, Despite Opposition to Electrification Policies in the U.S.
- Recorded surge in electric vehicle registrations during the initial half of the year
- Dubai-bound: Omega Seiki Mobility, an electric vehicle company from India, prepares for assembly establishment
