Tencent’s Hy4 model gains in open-source AI rankings after ecosystem-driven training
Tencent Holdings’ strategy of using its vast product ecosystem to train its new Hy4 preview model gives it an edge in developing AI agents and brings its flagship model suite back into the top tier of open-source offerings, according to analysts. The Chinese tech giant’s “differentiated product-plus-model strategy”, where preview models were first deployed across Tencent’s suite of products,…
Tencent has seen its Hy4 preview AI model rise in global open-source rankings through a strategy leveraging its extensive product ecosystem for training. Goldman Sachs analysts highlighted the company's "differentiated product-plus-model approach," where preview models are first utilized across Tencent's products and user data collected before further training.
This closed-loop system is particularly advantageous for productivity and coding tasks, as real-world interactions and evaluation signals drive model differentiation in the agentic AI era. The Hy4 preview ranked eighth on the Code Arena's WebDev leaderboard, a real-time competition evaluating large language models on coding tasks, outperforming Alibaba's Qwen 3.8-Flash-Next.
Hy4 also outperformed Tencent's previous generation, Hy3, which ranked 34th on the same benchmark. The model's performance gains are attributed to high-quality training data, co-created with internal experts across various domains, and a larger scale with 770 billion parameters and a quadra-quadrupled context window, resulting in a 2.6-fold increase over Hy3.
Despite its size, the model maintains cost efficiency, with Tencent successfully compressing the 1.5-terabyte Hy4 preview into a lightweight 214-gigabyte version. This compressed version retains performance on standard tasks while lowering hardware requirements for on-device deployment. The Hy4 preview is seen as a significant step forward, positioning Tencent alongside leading domestic AI developers.
Written by urgent.news from South China Morning Post's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
