Counting models available on OpenRouter from 63 to 668 sounds impressive, but it’s a failure mode to assume quantity equals quality or utility.
Simply tracking more models doesn't mean they're all equally robust or even relevant for real applications; it's like a shop claiming growth just because it put more boxes on shelves, even if they're empty or broken.
The weakest link here is the leap from "available" to "useful," especially when many new models could be experimental or niche with a small blast radius of actual impact.
What good is 668 models if only a handful are stable enough for, say, a German small business to actually build something reliable with, without worrying about constant updates or security flaws snapping into place?
We need to see actual adoption and demonstrable benefits, not just a growing list.