arXiv
In plain wordsA free website where scientists post research papers before, or instead of, formal publication. Most new AI research appears there first.
A free online server, run by Cornell University, on which researchers in computer science, mathematics, physics, statistics and neighboring fields post their own papers. Volunteer moderators screen submissions for subject fit and basic scholarly standards, but there is no peer review; each paper receives a permanent identifier such as arXiv:1706.03762 and appears in the daily listing of its subject category, for AI chiefly cs.LG (machine learning), cs.CL (computation and language), cs.CV (computer vision), cs.AI and stat.ML. Because AI moves faster than journal and conference cycles, arXiv became the field's de facto first channel of publication, and many influential documents, including technical reports for major language modelslarge language model, n.: A computer program trained on enormous amounts of text to predict the words that come next. At vast scale, that simple skill lets it chat, answer questions, translate, summarize and write code., exist only there. An arXiv citation therefore guarantees no review. In October 2025 its computer-science section stopped accepting review articles and position papers that had not already passed peer review, citing a flood of low-quality submissions.
It isn't peer reviewed yet — it's just on arXiv.
Origin
Started in August 1991 by physicist Paul Ginsparg at Los Alamos National Laboratory as an email service for high-energy-physics preprints, later reached on the web as xxx.lanl.gov. A Computing Research Repository section opened in 1998; the service was renamed arXiv in 1999 and moved with Ginsparg to Cornell University in 2001.
Related
- OpenReviewOpenReview, prop. n.: A website where conference papers are submitted and peer reviewed; for some conferences, anyone can read the reviews and the authors' replies.
- Papers with CodePapers with Code, prop. n.: A free website, run by Meta from 2019 until it closed in 2025, that linked AI research papers to their computer code and kept leaderboards of the best results.
- NeurIPSNeurIPS, prop. n.: The largest yearly scientific conference on artificial intelligence and machine learning, held each December, where many of the field's landmark papers were first presented.
- ICLRICLR, prop. n.: A major deep-learning conference, held every spring since 2013, where the reviews of each submitted paper are posted online for anyone to read.
- Attention Is All You NeedAttention Is All You Need, prop. n.: The 2017 Google research paper that introduced the transformer, the neural-network design behind ChatGPT, Claude, Gemini and nearly every other modern AI language system.
Mentioned in
ICMLICML, prop. n.: A leading machine learning conference, first held as a workshop in 1980, where researchers gather each summer to present new learning methods and theory., NeurIPSNeurIPS, prop. n.: The largest yearly scientific conference on artificial intelligence and machine learning, held each December, where many of the field's landmark papers were first presented., JMLRJMLR, prop. n.: A top machine learning journal, started in 2000 by researchers who wanted an alternative to commercial publishing, that costs nothing to read or to publish in., OpenReviewOpenReview, prop. n.: A website where conference papers are submitted and peer reviewed; for some conferences, anyone can read the reviews and the authors' replies., Papers with CodePapers with Code, prop. n.: A free website, run by Meta from 2019 until it closed in 2025, that linked AI research papers to their computer code and kept leaderboards of the best results.