Hy4 preview is a 770B-parameter model optimized for long-horizon software engineering, document-heavy office analysis, and scientific research tasks.
What the product does and how it is positioned
Hy4 preview represents a flagship generation of models from Tencent, optimized for long-horizon tasks and document-heavy professional workflows.
The model is designed to integrate with Tencent products and is available via open-source platforms to support diverse research and development needs.
Source-supported ways to use the product
Assists with understanding, planning, debugging, and verifying long-horizon development tasks, including front-end visual design.
Supports research questions in fields such as AI research, molecular dynamics, condensed matter physics, and pure mathematics.
The documented workflow, where available
Takes messy context spread across many files and converts it into shareable artifacts like documents, spreadsheets, and presentations.
Hy4 preview is built upon a 770B-parameter architecture with 49B active parameters. The development process focused on scaling model size, context length, and training data.
Checks to run with your own material and workflow
What was checked and when
Answers based on the source-checked product record
Hy4 preview is a 770B-parameter model that utilizes 49B active parameters to manage complex reasoning and large-scale data processing tasks.
The model supports a 1M-token context window, allowing it to process and synthesize information across extensive documents and long-horizon projects.
Hy4 preview is available through Tencent Cloud and OpenRouter, and it is also integrated into Tencent products such as CodeBuddy, WorkBuddy, Yuanbao, and ima.
The current version may spend more time than necessary when reasoning through complex tasks and has a documented tendency to over-verify its own work.
The model is optimized for long-horizon software engineering, document-heavy office work, game development, and scientific research tasks.
