papers posted to math arXiv per week in 2025-2026
Image from Daniel Litt (@littmath)

Nov 8, 2026 · 5:36 PM UTC

22
38
9
356
48,882
RelevantRecentLikes
Here's TCS and physics for comparison. TCS has really blown up; physics seems to have been increasing but not as explosively, though hep-th (which is pretty small) shows a substantial increase.
Image from Daniel Litt (@littmath)
Image from Daniel Litt (@littmath)
3
2
21
3,322
Is the implication here that slop is being posted? Or that a lot of good mathematics is happening faster?
2
1
658
There’s no implication!
1
4
566
If only we had a corresponding graph of papers *read*.
1
1
11
1,098
turn this one upside down
1
63
1,988
is there also data on the number of papers for the median researcher? i.e. is this growth being driven by everyone doing 2x the papers or like 10% spamming it like crazy?
1
4
517
Not sure, will think if there's a good way to see this.
1
7
1,410
View more replies
isn't Math.AP for PDEs? one would think AI would accelerate the submissions in PDEs more than others but in fact it is significantly less than others? do you have date on ALGTOP?
1
2
674
These are just the 5 biggest categories; all categories show more or less the same pattern. math.AT is a bit small for anything conclusive but there's clearly also a jump.
4
605
Has there been any change since they limited it to two submissions per month?
1
1
456
Probably too soon to really tell.
1
437
One possibility I see playing out in my own field is that people had a lot of papers on the back burner that they were slowly working through, and LLMs have allowed them to “clear their docket”.
4
188
are cross-lists included? if so some of the math and TCS jumps might be the same papers counted twice
1
61
math papers are reproducing faster than the theorems can be proven
1
148
contrary to the pessimistic takes, AI is leading to an explosion of new questions, new research opportunities. See for example the phantastic grass root follow up research surge kicked off by the recent "Z is diophantine in Q" result @ElliotGlazer which in turn was motivated by the latest OpenAI paper "dump". AI did not "nuke" the field but, to the contrary, dropped a flood of seeds into fertile ground now growing a vibrant jungle of new research driven not by the old pessimist AI haters but by a new generation of intellectually curious and courageous mathematicians who embrace AI.
1
154
How many papers on the edge got that last push by models and are finally seeing the light 🙌
1
2
115
Have you looked at whether withdrawal or revision rates have changed alongside the submission volume? Neither is necessarily a measure of quality, but I’d be curious whether those trends are changing too.
122
Thats crazy😅
2
418
It seems to have somewhat stabilized in math since arXiv announced limiting the new submissions to 2 per month (which I believe was a good decision).
4
141
same energy
Image from Alex Klotz (@AlexanderRKlotz)
1
239
the avg number of pages over time might also be interesting, given the current policies.
49
Math is dead already
1
81
"I'm not using AI, I'm just having a productive streak. Weather's nice today."
37
This explosive growth will slow and eventually halt once the excitement fades and math becomes too easy to be a good benchmark for AI. New platforms are also likely to emerge, since reading papers alone will become old-fashioned and no one will waste time reading them
2
127
... I'm not asking of you Just appreciate my thoughts, I don't want compliments Just read my words as well! What's wrong if I'm poor? What's my fault if I'm black? You can close all the doors I request you, read my work! -- Imran Islam
1
131
Load more replies