Tencent launches Hy4 with vision and computer use

Tencent announced its new large language model, Hy4, on November 4, 2026, describing it as ranking among the world's leading open-source AI models. The model adds visual understanding for the first time, including fine-grained image processing, chart analysis, and mathematical visual reasoning, along with computer-use abilities that let it interact with command-line and graphical interfaces. Tencent also trained it specifically on professional software such as Blender and CAD tools.

Tencent reported strong internal benchmark results, including a 83.18 average score and 96.15% pass rate on its WorkBuddy test, first place in nine of twelve scenarios in mainland China, and improved scores over the prior Hy4 Preview model on financial analysis, investment research, hallucination rate, and game development benchmarks. Hallucination rates reportedly fell from 12.95% to 7.2%.

Model weights and system cards are planned for release in the coming days, with a stated focus on cybersecurity and biological safety review. Hy4 is already available through Tencent Cloud TokenHub, WorkBuddy, CodeBuddy, Yuanbao, and other Tencent products, with API access via OpenRouter, Hermes, and OpenCode. Pricing is unchanged from Hy4 Preview at about $0.90 per million input tokens and $2.70 per million output tokens.

Why it matters

Hy4 adds vision and computer-use skills to Tencent's open-source model line, broadening what developers can build on it for free.

Source: Tencent