I Benchmarked Chinese vs US AI Models: The Numbers Don't Lie

I Benchmarked Chinese vs US AI Models: The Numbers Don't Lie

I've spent the last three months pulling API receipts and running side-by-side evals because I got tired of the discourse. Every LinkedIn thread I saw was either "China will eat US AI alive" or "Western models are years ahead" — neither claim had a spreadsheet attached. So I built one. What follows is everything I learned, with statistically ugly honesty about sample sizes and the kinds of correlations that survive a basic sanity check.

Let me walk you through what I found.

Why I Even Started Counting

My day job involves picking the cheapest model that doesn't make my pipeline look drunk. Six months ago I would have defaulted to GPT-4o without thinking. Then a contractor on my team pinged me about a Chinese model doing weirdly well on a long-context extraction task. I shrugged, ran a few prompts, and noticed my bill dropped by a factor I didn't believe at first.