I’m considering Perplexity for early-stage literature searches, where relevant work may be available as a preprint before it appears in a journal or conference. I’m not sure whether its results give preprints a fair chance of showing up, or mostly surface pages that are already widely cited and indexed.

For people using it in research, have you compared its academic results with searches in scholarly databases? I’d be interested in how it handles newer papers, working-paper versions, and cases where a preprint title differs slightly from the eventual publication.