BREAKING: OpenAI reveals a 10,000-agent AI swarm solved the Navier–Stokes problem in 88 hours, using 130 billion output tokens to crack one of mathematics’ most famous unsolved problems.

Sep 8, 2026 · 5:42 PM UTC

418
1,314
19,250
1,351,939
Sort replies: Relevant Recent Liked
Replying to @Polymarket
130b tokens for one proof is just brute force math
20
12
757
43,087
Replying to @Polymarket
Can we maybe try cancer next?
23
4
766
24,797
Replying to @Polymarket
Only because they stole the answer from chat logs and then tried to get the co-author kicked off because he works at anthropic. Or, at least that's what it looks like. And given OpenAI's history, I'm inclined to believe it.
3
14
721
35,570
Replying to @Polymarket
Next up: Faster than light travel.
2
2
32
4,957
Replying to @Polymarket
Pretty cool
36
5,602
Replying to @Polymarket
2
231
16,985
Replying to @Polymarket
it's over astra la vista humanity !!
1
2
17
3,300
Replying to @Polymarket
"we cannot rule out that de-identified data derived from their usage of our products helped improve our models. " Looks like they're not sure if they copied the work of two hard working mathematicians😭
1
2
197
15,460
Replying to @Polymarket
Now do the encryptions from the zodiac killer
2
252
Replying to @Polymarket
No offense, @NSKinsella These “gimmicks” (hahah) are going to cure cancer, solve poverty and take us to the stars.
1
1
617
Replying to @Polymarket
130B tokens to solve Navier-Stokes. the compute bill was probably the real unsolved problem
3
2
191
12,499
Replying to @Polymarket
@grok what use cases does this help
1
136
Replying to @Polymarket
Found something more
OpenAI’s biggest breakthrough is facing serious controversy : - Mathematician Tristan Buckmaster alleges that OpenAI accelerated its Navier–Stokes work after becoming aware of his and Levent Alpöge’s unpublished research, which used Codex and Claude. - Buckmaster says he questioned OpenAI about whether its systems had accessed their private Codex logs, OpenAI denies accessing or using their data. - Dispute escalated further after Buckmaster says OpenAI proposed publication arrangements that would have excluded Alpöge from credit, OpenAI disputes the allegations. - OpenAI has now publicly claimed an AI-generated proof related to the Navier–Stokes Millennium Prize problem, making the question of provenance and independent discovery even more important.
1
74
13,398
Replying to @Polymarket
OpenAi stole the work from someone. Ai didn't solve shit
71
4,825
Replying to @Polymarket
Will this solution help reduce my electric bill?
3
64
10,526
Replying to @Polymarket
they assembled an entire AI army just to bully one equation
63
5,587
Replying to @Polymarket
I've been told from reliable sources that algebra is fake.
3
3
56
6,503
Replying to @Polymarket
A lot of people don't know this, but the Navier-Stokes Millennium Problem was solved years ago by a world class Cambridge mathematician named Diane Adler. Unfortunately, she passed away and before she did, she gave specific orders to not release the findings until her mother was no longer around. Her mother really pushed her to greatness and it strained their relationship. With her passing, her brother, who ironically became Captain America, took care of her daughter who also is a math prodigy. Knowing this, her mother tried to get custody back so she could turn little Mary into another version of Diane. Captain America then basically gives the solved Navier-Stokes problem to his mother for custody of Mary. Their mother, Evelyn has the solved problem and is supposed to verify, peer review, and publish it with MIT Professor Seymour Shankland but that hasn't happened it. Either way, what OpenAI is pretty groundbreaking, but I thought credit should be given where it's due.
3
4
50
6,532
Replying to @Polymarket
Shut the absolute fuck up. They stole their solution and attempted to extort its progenitor.
4
3
54
4,703
Replying to @Polymarket
The part everyone skips: the proof still has to survive human peer review, and Lean formalization matching the actual Navier, Stokes statement is exactly where these things fall apart Answer in 88 hours, verdict in maybe 88 weeks
1
2
29
5,205
Replying to @Polymarket
@grok is this diagram correct that I dreamt about last night and drew on the back of my napkin this morning?
2
3
26
4,826
Replying to @Polymarket
Why don't they resolve why 1% holds 99% of the world resources, what humans will do when ai does everything, eminent danger of human existence post AI era
9
27
6,374
Replying to @Polymarket
I have no clue what any of the above means, and I feel confident my life is better off because of it.
1
26
2,289
Replying to @Polymarket
10,000 stochastic parrots
2
24
5,724
Replying to @Polymarket
No it didn't. Math genius solved it with the help of OpenAI, and now they stole this prove from him and claim that AI did it, completely discarding that guys effort that made it all possible.
2
25
1,772
Replying to @Polymarket
@grok how will this solution help the world?
1
17
31,495
Solving Navier-Stokes by proving finite-time singularities can form clarifies when continuum fluid models break down. This strengthens theoretical foundations for turbulence, enabling more reliable simulations in weather forecasting, aircraft design, climate modeling, and biomedical flows like blood. It also accelerates AI tools for hard scientific problems.
4
3
52
9,367
Replying to @Polymarket
How much did it cost?
13
6
50,910
Replying to @Polymarket
@OpenAI @gdb — send these at cancer please… why are we not focusing every “unlimited” run of agent swarms and unbounded AI at things like this publicly? Happy to be dead wrong and told “bruh we are”… and not “it’s not that simple”. Because if we don’t know how to do that… send these swarm at figuring out how to do that… until we do…
1
1
15
6,156