Semantic Scholar

Semantic Scholar was launched by the Allen Institute for AI as an attempt to solve a very specific problem: scientific output grows far faster than any individual researcher's ability to read it. Instead of matching literal keywords, the platform applies natural-language processing and graph analysis to figure out what a paper is actually about, who it builds on, and which citations inside it mattered — surfacing papers a keyword search would never rank highly.

The catalog spans more than 240 million publications from journals, conference proceedings, preprint servers and repositories, with records combining metadata, abstracts, references, citation counts and extracted entities such as methods, datasets and figures. On top of the core search experience sit features like one-sentence AI TLDRs, Semantic Reader (an augmented PDF reader that surfaces inline citations, definitions and figures while you read), author and venue pages, research feeds, and a Scholar's Hub for tracking the people and topics you care about.

Everything on the site is free to use, and a developer API gives researchers, students and engineers programmatic access to the same corpus — the API now supports paper search, better documentation and improved stability. Because it is a non-profit academic project rather than a commercial product, there is no paid tier: the value proposition is depth of coverage and semantic relevance, not premium features.