Tencent unleashes 770B-parameter Hy4

Tencent has released and open-sourced Hy4 preview, a new large language model designed for productivity tasks including coding, office work and scientific research.

The model has 770 billion total parameters, 49 billion active parameters and a context window exceeding 1 million tokens.

Hy4 preview is available through Tencent products including WorkBuddy, CodeBuddy, Yuanbao and ima, as well as through APIs via Tencent Cloud TokenHub and OpenRouter.

Tencent said the model will be available free on WorkBuddy and CodeBuddy for two weeks following launch.

The company said Hy4 preview was trained using data developed with experts across software engineering, gaming, finance and security.

In coding, the model is designed to handle long-context development, planning, debugging and validation. For office work, it supports data analysis and cross-document tasks involving documents, spreadsheets and presentations.

Tencent said Hy4 preview can also generate playable game prototypes from natural-language prompts and assist with scientific research in areas including AI, molecular dynamics, condensed-matter physics and mathematics.

In an internal blind evaluation covering 203 engineering tasks and 163 experts, Tencent said Hy4 preview scored 2.99 out of 4, compared with 2.94 for Kimi K3 and 2.92 for GLM-5.3.

The company also said the model helped optimize parts of its own development and inference system, increasing end-to-end throughput by 31.8 percent from its baseline.

API pricing starts at $0.834 per million input tokens and $2.501 per million output tokens, with cache hits priced at $0.042 per million tokens.

Tencent said additional models under the Hy4 series are expected to be released.