Before ResearchRabbit made visual discovery fashionable, Connected Papers (connectedpapers.com) was already doing it — and arguably doing it more elegantly. The premise is simple: feed it one paper you care about, by title, DOI, arXiv ID or URL, and it builds a single beautiful graph of the most similar research around it. In one screen you can see a field’s shape — its foundational works, its recent developments and the gaps between them.

What sets it apart is how it decides what’s similar. Rather than just following direct citations, Connected Papers uses co-citation and bibliographic coupling — analysing shared references — to surface conceptually related work even when two papers never cite each other. That’s a meaningfully different lens from ResearchRabbit’s citation-trail approach, and it routinely turns up papers a keyword search would never find.

How Connected Papers works

One seed, one graph

Enter a seed paper and Connected Papers analyses roughly 50,000 candidate papers to find the most relevant, then lays them out as an interactive graph. Each node is a paper; the layout positions similar work close together, so clusters reveal sub-fields at a glance. Click any node — or any entry in the side list — and a detail panel shows the paper’s information.

A graph you can read instantly

The visualisation is cleverly encoded so it communicates without explanation. Node size reflects a paper’s influence (citation count and centrality), so seminal works stand out as larger nodes. Colour represents publication year, letting you trace the evolution of an idea from foundational papers to the current frontier. Proximity reflects similarity. It’s the rare research tool you genuinely understand within seconds of your first graph.

Prior works, derivative works and multi-graphs

Alongside the main graph, Connected Papers surfaces “Prior Works” (the foundational ancestors a field is built on) and “Derivative Works” (the recent papers building on it) — a fast way to find both roots and frontier. You can also generate graphs from multiple seed papers at once to compare research areas, spot overlaps and explore the connections between different topics or approaches.

Built on the Semantic Scholar corpus

Connected Papers draws on the Semantic Scholar Paper Corpus — hundreds of millions of papers across every scientific field, from machine learning to biology to philosophy. It runs in the browser, offers extensions for arXiv, PubMed and other databases, and lets you export graphs as images or share them via embed links.

How Connected Papers scores

Overall score: 7.9 / 10, the average of the eight categories above.

Connected Papers pricing

Connected Papers runs a freemium model. The free tier gives you a limited number of graphs each month; paid plans simply remove that cap. Importantly, the features are identical across tiers — the only difference is how many graphs you can build.

Some listings quote pricing “from $6/month”. Confirm current rates on the official site, as the figure varies by source and billing cycle.

Where it falls short

The free tier caps your graphs

The handful of free graphs a month is plenty for occasional use, but if you’re running an intensive literature review you’ll hit the limit quickly and need a paid plan. It’s affordable, but it’s no longer the “completely free” tool some older write-ups still describe.

A snapshot, not a workspace

Connected Papers is built around the single graph, and that’s both its strength and its ceiling. It doesn’t offer the evolving, collection-based workflow that ResearchRabbit does — there’s less to organise, save and grow over a long project. Each graph is a brilliant snapshot of a moment, rather than a living map you build over weeks.

Discovery only

Like its visual-mapping peers, it finds papers but doesn’t help you digest them: no built-in reader, no annotation and no TLDR-style summaries. And because it sits on the Semantic Scholar corpus, it inherits that corpus’s coverage characteristics — strong across the sciences, with the usual caveats for completeness in a formal systematic review.

Pros and cons

Who should use Connected Papers?

Connected Papers is ideal when you want to understand a research area fast: a student starting a literature review, a researcher entering an unfamiliar field, or anyone who needs to find the foundational and frontier papers around a single key study. Its co-citation lens and instantly legible graphs make it the quickest way to get the lay of the land from one seed paper.

For sustained, organised reviews you’ll want a complement. ResearchRabbit offers evolving collections and ongoing recommendations; Semantic Scholar adds broad search and TLDR summaries; Scite tells you whether key findings hold up; and Consensus or Elicit handle synthesis and extraction. Litmaps is the other obvious visual-mapping rival worth comparing.

Verdict

Connected Papers is a small, focused tool that does one thing exceptionally well. Its co-citation and bibliographic-coupling approach genuinely surfaces work other methods miss, and its graphs are the clearest and most beautiful in the category — you understand a field within seconds of generating one. As a way to map the territory around a key paper, nothing is quicker or more elegant.

It lands just behind ResearchRabbit because it’s a snapshot rather than a workspace: the free graph cap, the single-graph focus and the absence of a reader or summaries mean it shines brightest as one part of a wider toolkit. Used that way — and it’s cheap enough to be — it’s a delight. Score: 7.9/10.

Frequently asked questions