arXiv's two-a-month submission cap has been in force since 1 October

arXiv’s rate limit is already in force. The blog post Fair Moderation, Equitable Access, and AI, by Kat Boboris and dated 1 October 2026, says: “As of October 1st, 2026, arXiv is instituting an updated rate limiting policy across all submitters and categories.” 404 Media wrote about that policy on 5 October. The effective date in the primary post is 1 October.

The limits, as the blog states them

arXiv “now limits submitters to up to two submissions per calendar month, with a limit of three total active submissions at any given time.” The FAQ says the limit “applies to all submissions, across all categories.”

A rejection counts. The FAQ says the limit “is for submissions, not announced papers, because it is submissions that consume moderator time,” and that two papers in a month, accepted or rejected, mean waiting until the next month.

Holds carried from last month do not count against the new month’s two submissions. They do count toward the three active submissions. The blog says that active cap has stood since 2024. Deleting a paper before it is announced consumes neither limit. A submission counts only against the person who sends it, not against co-authors.

The post does not describe an exemption or an appeal.

The volume figures arXiv gives

The same post gives three September totals. “In September of 2016, arXiv received 9,869 submissions. In September of 2024, arXiv received 20,569 submissions. This September, arXiv received 40,363 submissions, which in turn generated almost 9,000 support tickets for arXiv staff and moderators.” It says that in the past two years submissions have doubled. A separate line says increases over the past two years in cs.AI, “over 6X,” illustrate the rate, and that other categories have seen submissions double over the same period.

404 Media repeats those three September totals and says submissions have sextupled in computer science. The blog’s “over 6X” line is specifically cs.AI.

What arXiv says is driving the cap

The blog says arXiv still permits AI “as a tool in support of their research as long as it is disclosed and the research meets arXiv’s standards of scholarly interest and advances the field of research.” It says many new submissions do not meet that bar: more “thin papers of narrow scope,” “salami” papers that split one work into several submissions, and “dense, AI-written papers.”

404 Media quotes Thomas Dietterich, described there as an Oregon State University professor emeritus and chair of arXiv’s editorial advisory council, “in a statement”: “The heart of arXiv’s process for detecting and rejecting low-quality papers is our many volunteer moderators.” The same quotation continues that moderators “donate their time and expertise every day,” and that “a relatively small proportion of authors are submitting a large number of low-quality papers and consuming a disproportionate fraction of the moderators’ time.” Those sentences also appear on the arXiv blog, set off as a quotation under his name and title. The blog says the update is meant to spread moderator time and to protect the corpus, and that arXiv will “monitor the effect of this update” and “modify as needed.”

What a submitter can do with this

Count submissions, not only announced papers. Two in a calendar month, accepted or rejected, ends the month. Three active submissions is a separate cap, including holds carried over from last month. A delete-before-announcement does not consume either count. Nothing on the 1 October post offers a waiver.