Tencent Holdings Ltd. has released Hy4 Preview, a new foundation model, according to Bloomberg. Techmeme’s Bloomberg-linked summary says the model has 770 billion parameters and a 1-million-token context window. The core claim is competitive: Tencent says Hy4 Preview outperforms models from Z.AI Co. and Moonshot AI in internal tests. The provided items do not include independent benchmark results, test names, pricing, availability terms, or deployment details, so the performance claim should be treated as Tencent’s assertion rather than an externally verified ranking. Bloomberg frames the release as Tencent’s bid to claim a place among China’s top artificial intelligence players. That matters because Z.AI and Moonshot are among the named rivals Tencent is directly comparing against, and the company is using model scale, context length and internal benchmark results to position Hy4 Preview in that competitive set. The most concrete specifications in the cluster are the model name, parameter count and context window. The most important unresolved question is whether external users or third-party evaluators can reproduce Tencent’s claimed advantage over Z.AI and Moonshot models outside Tencent’s own internal tests. Who benefits: Tencent benefits if Hy4 Preview strengthens its position with customers and developers evaluating Chinese foundation models. Users needing long-context models may have another option to assess, subject to real-world performance checks. Who's exposed: Z.AI and Moonshot are the directly named competitors in Tencent’s internal comparisons. The exposure depends on whether Tencent’s claimed benchmark advantage holds up outside internal testing.