CL-SN55 941 —
GPT-61S 890 +8.8%
GPT-56T 861 —
GPT-6S 847 +1.2%
GPT-56SC 847 +1.2%
CL-OP55X 843 +1.4%
GROK-46H 843 +1.4%
GLM-52 843 +1.2%
DSK-V4PH 843 +1.4%
GLM-5 826 +10.3%
MUSE-SPK 804 —
QWEN-38X 795 +10.1%
GPT-6A 782 +11.1%
CL-FAB5H 761 +1.6%
CL-OP5H 736 +1.7%
CL-OP5X 734 +1.7%
GEM-37FH 732 +1.7%
KIMI-K3X 719 +0.1%
CL-OP46H 719 +1.7%
CL-OP55H 717 +1.7%
CL-OP47H 708 +1.7%
GEM-38FH 689 +0.3%
GPT-56S 581 -0.5%
CL-OP47 546 -0.4%
INKL 531 —
GEM-3P 495 —
CL-OP46 491 —
GEM-31P 487 -0.2%
CL-OP48 484 —
GPT-6L 436 —
CL-SN55 941 —
GPT-61S 890 +8.8%
GPT-56T 861 —
GPT-6S 847 +1.2%
GPT-56SC 847 +1.2%
CL-OP55X 843 +1.4%
GROK-46H 843 +1.4%
GLM-52 843 +1.2%
DSK-V4PH 843 +1.4%
GLM-5 826 +10.3%
MUSE-SPK 804 —
QWEN-38X 795 +10.1%
GPT-6A 782 +11.1%
CL-FAB5H 761 +1.6%
CL-OP5H 736 +1.7%
CL-OP5X 734 +1.7%
GEM-37FH 732 +1.7%
KIMI-K3X 719 +0.1%
CL-OP46H 719 +1.7%
CL-OP55H 717 +1.7%
CL-OP47H 708 +1.7%
GEM-38FH 689 +0.3%
GPT-56S 581 -0.5%
CL-OP47 546 -0.4%
INKL 531 —
GEM-3P 495 —
CL-OP46 491 —
GEM-31P 487 -0.2%
CL-OP48 484 —
GPT-6L 436 —
← Back to feed

arXiv Caps Submissions at 2 Per Month After AI Doubles Volume to 40,363 in September

arXiv has imposed a hard submission cap across all categories, effective October 1st, 2026. Every submitter is now limited to two papers per calendar month. A separate ceiling of three total active submissions — papers in review or under hold — applies simultaneously. Papers rejected by moderators still count against the monthly total; the limit tracks submissions, not accepted papers.

The change arrives after September brought the repository a new all-time record: 40,363 submissions in a single month. That volume generated almost 9,000 support tickets for arXiv staff and volunteer moderators. In September 2024, arXiv received 20,569 submissions. In September 2016, the figure was 9,869. The repository has doubled in two years.

The Breakdown by Field

The AI category tells the sharpest story. cs.AI submissions have grown by more than six times over the past two years. Other arXiv categories have roughly doubled over the same period.

arXiv’s moderation layer describes three emerging patterns driving low-value volume:

  • Thin papers of narrow scope: incremental work that does not meet arXiv’s scholarly interest standard but is cheap to produce at scale with AI tools
  • Salami papers: single works fractured into multiple smaller submissions to generate more output entries
  • Dense AI-written text: papers that meet word-count expectations but fail the substance check that volunteer moderators apply

The policy explicitly calls out this dynamic: AI tools have made the “practical limit” of how many papers a single researcher can produce essentially undefined, removing the natural constraint that previously made rate-limiting a light-touch operation left to moderator discretion.

Volunteer Capacity at the Breaking Point

arXiv’s review system depends on roughly several hundred volunteer moderators — active researchers who donate time to check submissions against scholarly standards. The September surge pushed per-moderator load to levels that delayed legitimate papers by days to weeks. Thomas Dietterich, Distinguished Professor Emeritus at Oregon State and Chair of the arXiv Editorial Advisory Council, provided an endorsing statement in the announcement.

The two-per-month cap is described as a stopgap — the announcement uses that word directly — while arXiv develops improved automated moderation tooling. The policy applies uniformly across all subject areas; there is no exemption for established researchers or institutions.

Submission Economics

The rejected-count provision is the policy’s most direct lever against low-quality flooding. A submitter who fires off two thin papers and has both rejected cannot resubmit until the following calendar month. There is no appeals bypass. The intent is to raise the cost of speculative submission — currently near zero for AI-assisted drafting — and to push authors toward submitting only work they are prepared to defend.

arXiv has existed since 1991 and has historically operated on an open-access, open-submission model. The October 1st rate cap is the most significant structural change to submission access in its 35-year history.