Hunyuan Hy3 Preview Tops OpenRouter's Global LLM API Usage Rankings

A Chinese LLM Tops Global API Usage Rankings for the First Time

On April 29, OpenRouter, a globally renowned LLM API aggregation platform, released its latest model usage rankings. Tencent's new Hunyuan Hy3 preview model surged to the top of the overall leaderboard, becoming the most-called large language model among developers worldwide. This marks the first time a Chinese-made LLM has claimed the number-one spot on the platform's comprehensive rankings, drawing widespread attention across the industry.

Hy3 Preview Delivers Strong Performance Across Multiple Scenarios

As one of the world's most prominent LLM API routing platforms, OpenRouter aggregates hundreds of models from major providers including OpenAI, Anthropic, Google, and Meta. Its usage rankings reflect, to a significant degree, the real-world preferences of global developers for different models.

According to the latest ranking data, Tencent's Hunyuan Hy3 preview not only ranked first in overall API call volume but also delivered impressive results in specific use-case categories:

• Tool Use scenarios: Ranked first
• Coding scenarios: Ranked second

Tool-use capability is a critical metric for evaluating a large model's practicality in agent-based applications, directly determining whether a model can efficiently interact with external APIs, databases, search engines, and other tools. Hy3 preview's top ranking in this dimension demonstrates strong developer endorsement for its performance in cutting-edge applications such as AI agents and automated workflows.

In terms of coding ability, Hy3 preview ranked second, further validating its competitiveness in core technical tasks such as code generation, debugging, and comprehension.

Chinese LLMs Accelerate Global Expansion

Hunyuan Hy3 preview's rise to the top of OpenRouter reflects the growing influence of Chinese-made large models in the global developer market. Over the past year, Chinese models represented by DeepSeek, Qwen, and Hunyuan have been listed on major overseas platforms one after another, gradually transitioning from "catching up technologically" to "penetrating ecosystems."

Notably, OpenRouter's usage rankings are driven by the actual behavior of real developers. Compared with traditional benchmark leaderboards, this data more accurately reflects a model's popularity in real-world production environments. Hy3 preview's ability to stand out among numerous world-class models suggests it holds strong competitive advantages across comprehensive dimensions including cost-effectiveness, response speed, and task completion quality.

Tencent's Hunyuan team has noticeably accelerated its model iteration recently, from upgrading its MoE architecture to optimizing multi-scenario capabilities, continuously improving model practicality and developer experience. The "preview" label in the Hy3 preview name indicates this is still a preview version, suggesting that the official release could deliver even greater performance improvements.

Outlook: From API Usage Volume to Ecosystem Moats

Topping the API usage rankings is undoubtedly a milestone achievement, but sustaining a leading position ultimately depends on continuous model iteration capabilities and the depth of ecosystem development. As global LLM competition enters the second half — where applications reign supreme — real-world developer usage data is becoming the core metric for measuring model competitiveness.

For Chinese LLM providers, performance on overseas platforms like OpenRouter serves not only as proof of technical strength but also as a critical gateway to the global developer market. Hunyuan Hy3 preview's rise to the top may further boost the confidence and momentum behind Chinese large models expanding internationally.