They've fallen behind in testing capacity, like many others. You have to go by hospitalizations and confirmed deaths because those are the only stats likely not to be underreported too badly.
Maybe you're being fooled by the lack of reported deaths in the weekends in Sweden? It does mess up the curve quite a bit. But it's been like that every weekend if you look at daily deaths. And it makes the log graph look less exponential than it is
Also the log graph is a day behind the reported numbers, so tomorrow you'll see a sharp rise as they're at 1200 deaths now, making it clear that it's indeed exponential
You are right though, given the numbers there's decent likelihood that it's stagnating. But there's also a decent risk that it's still exponential