According to recent OpenRouter usage data, Xiaomi’s MiMo model has moved into the global number two position for AI model usage, behind only DeepSeek-V4-Flash. The data, cited by 36Kr based on OpenRouter’s weekly figures, shows the top five models in usage are all Chinese.
For the AI model market, this is a significant data point. The concentration at the top of the usage charts is Chinese, and the open-source model space is no longer competing for second place — it is setting the pace.
What the Numbers Show
The data is from OpenRouter, which aggregates usage across AI services. It is one of the better signals for real model usage at scale, since OpenRouter sees a significant share of API traffic.
The numbers, for the week of late July through early August 2026: Chinese AI models collectively accounted for 28.13 trillion tokens, with the top five models all Chinese. DeepSeek-V4-Flash was first at 7.22 trillion, Xiaomi MiMo was second at 6.3 trillion, Tencent Hy3 third at 4.82 trillion, DeepSeek-V4-Pro fourth at 3.28 trillion, and Zhipu GLM-5.2 fifth at 2.89 trillion.
The concentration of Chinese models at the top is the most striking data point. It is not that one model dominates; it is that the leaderboard is essentially Chinese across the top.
Why This Matters
For the global AI market, the data is a clear signal: Chinese open-source models are setting the pace in real-world usage.
A year ago, the top usage charts were more mixed, with US models in strong positions. The shift to Chinese dominance is the result of sustained investment, competitive model releases, and adoption in the Chinese developer ecosystem.
For developers and businesses, the practical implication is that the open-source model space has real choices now, and the top options are increasingly Chinese. The competitive landscape has shifted, and that affects pricing, capability, and ecosystem for everyone.
For observers of the AI industry, the data confirms what other signals have suggested: the open-source model space is strong, and Chinese models are leading it. The right takeaway is that the field is competitive and the options are real.
What to Watch
A few things will determine how the data evolves.
Continued release cadence. Chinese labs have been releasing major model updates frequently. If that pace continues, the top of the chart will keep shifting.
Ecosystem development. Usage rankings reflect developer adoption, which is influenced by tooling, documentation, and ecosystem. Watching how the ecosystems develop is the real story.
Comparison with proprietary. The interesting question is how proprietary models like Claude and GPT compare to open-source alternatives in actual usage. The data here is one data point, and the picture is dynamic.
Bottom Line
Xiaomi MiMo’s rise to global number two is a meaningful data point for the AI model market. The concentration of Chinese models at the top of usage charts is the real story.
For developers, the practical takeaway is that the open-source model space is competitive, and the top options are real. For businesses evaluating models, the data suggests that open-source is no longer a secondary choice.
The honest take: the data is one signal among many, and the model space changes fast. Xiaomi MiMo at number two is meaningful today, but the chart will look different in a few months. Watch the trend, not the snapshot.