Tencent has released Hy4 preview, a next-generation large language model (LLM) with 770 billion total parameters, 49 billion active parameters, and a context window exceeding 1 million tokens, targeting demanding workloads in coding, office productivity and scientific research.
The model is now available as an open-source AI model and can be accessed globally through Tencent’s WorkBuddy, CodeBuddy, Yuanbao and ima products. Developers can also connect to Hy4 preview through APIs via Tencent Cloud TokenHub and OpenRouter.
Hy4 preview was designed to handle complex, long-context workflows. In software development, it offers enhanced capabilities for code understanding, planning, debugging and validation, while also supporting front-end development and interaction improvements.
The model is also optimized for data analysis, cross-document collaboration and document generation, covering workflows involving information processing, spreadsheets and presentations. In game development, Tencent says Hy4 preview can generate playable prototypes from natural-language prompts and interact with game engines.
One of the model’s notable features is its involvement in its own development. Hy4 preview participated in automated optimization of training methods, data strategies, evaluation frameworks and low-level operators, creating an early-stage recursive self-improvement loop.
Tencent also said the model autonomously identified inference bottlenecks and optimized areas such as operator fusion and communication. These changes increased end-to-end inference throughput by 31.8% compared with the baseline.
In an internal blind evaluation involving 163 experts and 203 engineering tasks, Hy4 preview achieved an average score of 2.99 out of 4.00, compared with 2.92 for GLM-5.3 and 2.94 for Kimi K3.
API pricing is set at US$0.834 per million input tokens, US$2.501 per million output tokens, and US$0.042 per million cached tokens. Tencent said the next batch of Hy4 models is expected to roll out soon.
