About SummarizePaper

What you can do here

It started in 2023 as an arXiv summarizer. It now covers every step of research, from reading a paper to presenting it:

All the tools, described · Illustrated guides with real examples · Your first paper, step by step

Where the papers come from

Papers come from arXiv, bioRxiv, medRxiv, HAL and Europe PMC, and from journals through their DOI, with metadata from OpenAlex, Crossref and Semantic Scholar. We read the full text only when the license of the paper allows it, and the abstract otherwise; each page says which one was used. Summaries can be read in 16 languages.

We thank arXiv for use of its open access interoperability, and OpenAlex, Crossref, Semantic Scholar and the preprint servers for their open data.

How we keep the AI honest

Summaries, answers and letters are written by language models (OpenAI). They can be wrong, so we make them easy to check:

• chat answers and key findings show the passages of the paper they rely on, which open in place;
• the numbers of the weekly letters and the quotes of the press kits are checked against the papers before publication;
• when only the abstract could be read, the page says so instead of guessing the rest.

Free and Premium

Summaries, chat, collections, the daily digest, rising topics, the weekly letters and one saved search are free, within daily limits. Premium and Premium+ open the tools that read the most for you and fund the service; they can be cancelled in one click. See the offers

Who is behind it

SummarizePaper is created and developed by Quentin Kral, astronomer at the Paris Observatory, who wants everyone, researchers or not, to benefit from research findings and be able to check information at its source. It is published by CarbonFreeConf, a French company, and its code is open source (MIT license).

More about the author · Source code · Legal Notice

Get in touch

A question, a bug, an idea of a tool, or a partnership? We read every message. Contact us