Misleading headline.
For six consecutive weeks, Chinese AI models have outpaced their American counterparts in total API call volume on OpenRouter, one of the largest AI model routing platforms, signaling a decisive shift in global developer adoption patterns.
- It’s from openrouter. Most us model users don’t go through Openrouter (most chinese probably don’t either, but the ratio is probably higher for us).
- Also like half of openrouter’s volume if not more is handled by western inference providers since the model is open
- Openrouter rankings tend to be skewed by discounts and free models (like Minimax m3 is on a 50% discount) , both on openrouter or through apps like Kilo code which route through openrouter then provide a discount
I see little reason to expect that OpenRouter isn’t a representative sample. Your argument is based on the premise that people using OpenRouter behave substantively differently from other users so the distribution there wouldn’t be a good sample. But, what’s that argument based on? Why would users who pay for direct API access behave in a different fashion and prefer paying an order of magnitude more for models with comparable capability from American companies?
because
- as I just said “Openrouter rankings tend to be skewed by discounts and free models (like Minimax m3 is on a 50% discount) , both on openrouter or through apps like Kilo code which route through openrouter then provide a discount”
- also as I said “Most us model users don’t go through Openrouter (most chinese probably don’t either, but the ratio is probably higher for us).”
There’s several reasons for this. Openrouter providers easy provider switching and model discoverability in exchange for a 5% markup and some latency. Western models don’t have many providers anyway, their users are generally not exploring other models as much so both those benefits go away. Companies, which (outside of china) overwhelmingly prefer going to providers themselves and doing enterprise level contracts instead of going through openrouter. They also prefer western models. Most open model usage comes through individuals trying to reduce costs and/or explore other models, which are going to be overrepresented on openrouter.
My point was that there’s no reason to expect that the distribution is different outside OpenRouter. Also, companies absolutely do not prefer western models. https://www.youtube.com/watch?v=9baDOfwUzHQ
In fact, here’s an interview from Airbnb CEO plainly saying they prefer Qwen to ChatGPT:
Cursor’s Composer 2 is based of GLM https://news.ycombinator.com/item?id=47452404
and here’s an article from just a few days ago stating that US companies are increasingly using DeepSeek https://finance.yahoo.com/sectors/technology/articles/many-us-tech-firms-turning-202000738.html
There’s zero evidence for your assertion that Chinese models are overrepresented on OpenRouter. The reality is that these are the models everybody is using right now, and there’s a niche market for overpriced American models.
