Agi 2029 - AI Insider / Reporter as featured in Axios • The Information • NYT • Techcrunch

Pinned Tweet
🚨 OpenAI’s unreleased AI model has reportedly proved the quasi-Riemann hypothesis with Lean formalization released alongside it. If confirmed, this would be a HISTORIC breakthrough in number theory and a landmark moment for AI. The claim tackles mathematics deeply connected to how prime numbers are distributed. It would establish much stronger limits on how far prime counts can deviate from their expected pattern. The paper also derives consequences for longstanding number-theory problems and guaranteed deterministic algorithms for finding square roots modulo a prime.
23
50
876
66,802
Keep an eye on Arena for Kimi this week
7
1
97
15,315
Prime intellect will solve continuous learning and we will be freed
Replying to @willcb
the final algorithm is something i’m calling “agentic gradient” which is where an agent can just do whatever it wants to its harness or weights the meta-RL problem is getting it to actually work, and there are many many bridges to cross along the way
3
2
127
10,336
Tomorrow
39
6
646
140,006
I dug into the GPT-6.5 tease on the opensilol domain. The site really does hide “6.5” and a Nov 1 countdown in its HTML. But a Sept 23 archive called the same domain an “independent OpenAI parody” for sale. I’d take it with a grain of salt, but who knows.
WE MIGHT HAVE JUST GOTTEN THE FIRST SIGN OF GPT-6.5! The domain opensi.lol has a hidden "6.5" in its source code and a countdown to November 1, 10AM SF time. That's exactly Day 28 of Tibo's 28 days. Either this is the biggest troll of the week, or someone knows something we don't.
24
6
207
22,896
I agree with gremlin it’s most likely a troll, Nov 1st in a Sunday.
20
2,932
OpenAI models are apparently REALLY good at modernizing legacy software (which still power a large swath of the global financial services industry) OpenAI’s latest COBOL migration benchmark shows GPT-6.1 Sol reaching roughly 97% accuracy at just $0.50 per graded trial, while GPT-6 Sol needs around $2 to reach similar performance.
25
24
439
24,253
You can tell your Dot to contact every phone carrier’s customer support team to find you the best deal on a new phone, btw.
3
54
5,447
I had the craziest experience today. I was at the bank making a deposit, and there was this 50-year-old man with a PS5 shirt next to me. As the bank tellers went back to do a double-verified count of my deposit, the guy looks at me and says, "Have you seen people use AI to put all of those Sony games on a browser?" I immediately became wide-eyed. I thought I would be assassinated right then and there. Out of all the people in the world around us, he chose to look at me and specifically ask about AI enabling games to be ported to the web, or aiding in games being ported to the web. I said, "Yeah, it's pretty cool." He said, "Yeah, they don't want to sell us physical media, so they had it coming." I just nodded. It felt like a movie scene of people happy that we're entering the singularity.
36
25
861
36,049
Everyone I know with early access to Fable 5.5 said it's another step change in game development. I made a prediction earlier this year that gaming companies would cite AI as a concern in their shareholder meetings. So far, we've seen them speak about AI in their shareholder meetings, but more about transitioning towards AI (i.e. Epic Games). Really excited to see what I'll be able to cook up with Fable 5.5.
22
26
804
28,605
NO OPENAI WE NEED OUR ASTRA KILLER
GPT-6.1 Terra is in a Microsoft PR Microsoft updated 18 evaluation configs from GPT-5.6 Terra to GPT-6.1 Terra Wouldn't be surprised if we get this soon
10
6
355
11,438
I know I speak about AI, but I need you guys to know this is culturally the most important event of my life
Jeff Grubb says Disney is in talks with Bungie and DICE about a possible new Star Wars Battlefront game. The talks are reportedly about finding a studio to develop Battlefront 3. DICE has already worked on the previous Battlefront games, while Bungie has experience making large-scale first-person shooters.
13
1
117
5,195
8
1,600
Skyfall AI just introduced SKY-001, what they’re calling the first Engineering World Model and they’re claiming it outperforms GPT-6 Astra on CAD and BIM engineering tasks. their model is 51-125x faster and over 150-210x cheaper than the frontier models they tested, while achieving higher intersection-over-union scores on engineering benchmarks!
Introducing SKY-001, the first Engineering World Model. On CAD and BIM tasks, it beats GPT-6 Astra and every frontier model we tested: -> 51-125x faster - > 150-210x cheaper -> Higher IOU than all frontier models SKY-001 lets you dream a physical object before you build it. 18 months of work. Built by the team out of Maluuba's deep learning lab (acq. by @MSRResearch). The first ‘AI General Engineer’ will be the backbone for Autonomous Manufacturing.
Paid partnership (ad)
5
7
130
10,164
How did they achieve this pareto optimality??
32
9
526
46,795
Last week you may have caught the Space Bunny = minmax 3.1 post! 1 down / 1 to go.
Chinese Rumor Mill: GLM 5.5 - next week/late October Space Bunny Alpha - MinMax model I try to refrain from sharing news that isn't explicitly concrete or that I can't be sure of. However, the informants who told me these specific rumors are well-connected with both ZAI and MinMax. So, I wanted to put my hat in the ring on the rumor mill. We will have some more concrete stuff soon.
3
1
62
5,460
Minmax 3.1 this week.
Chinese Rumor Mill: GLM 5.5 - next week/late October Space Bunny Alpha - MinMax model I try to refrain from sharing news that isn't explicitly concrete or that I can't be sure of. However, the informants who told me these specific rumors are well-connected with both ZAI and MinMax. So, I wanted to put my hat in the ring on the rumor mill. We will have some more concrete stuff soon.
3
1
49
4,441
I think the best way to describe OCD to an AI bro is that we feel the dire necessity to mech interp our own reasoning traces, or our polysemantic neurons start screaming.
5
56
6,047
Me and @scaling01 at the function
5
2,055