Chinese tech giant Tencent has unveiled a new version of its Hy4 language model, significantly expanding its functionality. The key difference is the introduction of visual analysis and the ability to control a computer — both via the command line and through a graphical interface.

Hy4 is positioned as a full-fledged AI agent capable of interacting with professional software. In particular, the model demonstrates work with tools such as Blender and CAD systems, performing complex multi-step tasks, analyzing graphs and images. This takes the LLM beyond traditional text scenarios, turning it into a universal digital assistant for specialists of various profiles.

Expanding Applied Capabilities

The developer announced a noticeable improvement in the model's results in a number of applied areas — financial analysis, programming, and scientific research. Such improvements indicate Tencent's desire to establish itself in the corporate AI solutions segment, where accuracy and depth of analysis are critically important.

In the coming days, the company plans to open the model's weights and publish system cards, which will ensure transparency of the architecture and allow the research community to study Hy4's capabilities more deeply. Opening the weights could intensify competition in the open LLM market and accelerate the development of applied agentic solutions.

Cryptalist Analysis: The release of Hy4 with computer control functions reflects the general trend of transition from chatbots to autonomous AI agents capable of performing actions in real applications. Opening the model's weights could increase pressure on competitors in the open LLM segment and stimulate the growth of the tool ecosystem around agentic scenarios. For the market, this is a signal of further integration of AI into professional workflows, which forms a long-term demand for computing infrastructure and corresponding crypto and blockchain solutions.