Tencent Releases Hy4 Large Language Model

Tencent announced the release of Tencent Hy4 on November 4, 2026, marking the entry of its latest-generation large language model into the open-source ecosystem. The model introduces visual understanding and computer-use capabilities for the first time, supporting interaction with command-line interfaces and graphical user interfaces.

Hy4 adds training on professional software like Blender and CAD to handle long-horizon tasks. Evaluation benchmarks on Tencent WorkBuddy show a performance score of 83.18 and a pass rate of 96.15 percent. In a financial evaluation involving 213 professionals across 579 tasks, Hy4 recorded an evaluation score of 3.04, compared to 2.97 for DeepSeek-V4.1-Flash-1017, 2.68 for Kimi K3.1 Pro, and 2.68 for GLM-5.3. Tencent reported that Hy4 reduced its hallucination rate from 12.95% in the Hy4 Preview to 7.2%, while errors from unsupported assumptions fell from 23.1% to 5.6%.

In game development benchmarks, Hy4 improved by 19.1 percent in GameDev and 9.8 percent in GameLogic compared to the Hy4 Preview. The model also incorporates defenses focused on proof-of-concept development and vulnerability remediation while excluding offensive exploitation materials from its training dataset.

Hy4 maintains pricing of approximately US$0.90 per million input tokens, US$2.70 per million output tokens, and US$0.045 per million cached input tokens. From November 4 to 15, WorkBuddy and CodeBuddy are available for free. Additionally, the model is available on Tencent Cloud TokenHub at 50% off standard pricing through December 31, 2026, with Hy4 Flash scheduled for release soon.