Indeed all that matters are individual papers, both for evaluation of careers but also for the perception (bias, really) that one develops of a journal.
A colleague of mine had the most thorough and toughest review process ever with a submission to PLoS ONE, but that was about a decade ago if not more.
I've published there twice too, and while the first time (2012) the editor only secured one review and it was as mild as it gets, the second time (2022) we got two and they were thorough and insightful. By this (dramatically undersampled) trend alone the journal has improved. Counteracting this, the emails I keep getting from editors requesting that I review papers well outside my field and that frankly look like the kind of manuscripts that should never be published in the first place suggests the journal has changed.
I loved the idea of PLoS ONE when it came out: it's pretty much what eLife is doing nowadays with Reviewed Preprints. But these initiatives require competent editors that care deeply, and reviewers that put in the time and effort. Doing so for for-profit journals like Cell Reports or Sci Adv is, in principle, harder: free labour for a company. I mean why would one ever do that.
