pipette
ESEspañol

Open data

Everything Pipette produces is public. Each day adds about two thousand papers with the labels shown on the site, including every probability. Use it for research, teaching, dashboards or your own feed.

10,613 papers since 2026-09-21, updated every morning.

Files

/data/v1/index.json
List of days with counts
/data/v1/days/2026-09-27/edition.json
One day's edition, full records
/data/v1/days/2026-09-27/index.json
Every paper of a day, compact records
/data/v1/p/<id>.json
One paper's full record
/data/v1/method.json
The exact questions and ranking rule
/data/v1/search/recent.json
Titles of the last two weeks

What each paper record contains

id, title, authors, venue, date, url, pdf, doi
Identifiers and links to the original
status, published
Preprint or peer-reviewed journal, and the published version when known
field, topic
One of 13 fields and about 110 topics
sentences
The abstract split into sentences (open licences only)
j.key, j.caveat
Index of the sentence with the main result, and of the admitted limitation
j.kind, j.evidence
Kind of contribution and evidence, with the full probability distribution
j.appeal, j.advance, j.level
Calibrated scores for breadth of interest, claimed advance and reading level
j.hype, j.practical
Probability of overclaiming, and of direct real-world impact
rank, pick
Pipette's ranking score and whether it made the daily edition

Licence

Pipette's labels, rankings and editions are dedicated to the public domain (CC0 1.0). Titles and abstracts keep the licence of their source: arXiv metadata is CC0, bioRxiv and medRxiv abstracts carry the licence chosen by their authors, and journal metadata comes from OpenAlex (CC0). Closed-licence journal abstracts are not included.

Please credit “Pipette by Ferced (pipette.day)” when you can. It is not required.

How the labels are made