Chinese AI models have reached parity with top US counterparts, according to a recent analysis. Kimi K3 and GLM-5.3 now perform near the top of almost every broad, demanding evaluation, handling long knowledge tasks, multi-step coding, and tool coordination more reliably than their predecessors. The gap, often cited as a few months, has shrunk enough to become an investor concern. According to the Wall Street Journal, Anthropic is facing questions ahead of its IPO about its remaining lead. Below that tier, the field is dominated by open, cheaper Chinese models. Investors worry that raw model performance alone can no longer guarantee business success. Whatever a model can do today, a freely downloadable one can replicate in a few months.
The shift in the AI landscape has led to two accusations: Chinese labs allegedly used Western models as teachers through distillation, and they allegedly tuned models for strong benchmark scores without broader capabilities, a practice known as benchmaxxing. While the American lead hasn't disappeared, it has retreated to a few, ever-narrower areas of the so-called frontier, the leading edge of what's technically possible. Where that lead still sits, and what it's worth economically, is the first question of this issue. The second follows from it. If the model alone can't make a clear difference anymore, what does the lead rest on? Our thesis: less and less on the model, and more and more on the overall system around it, meaning the system where ongoing work produces the next models.
Source: thedecoder
