
Math from @IUIndianapolis Interested in economics, history, & energy policy
Indianapolis, IN
Joined April 2021
- Tweets2.5K
- Following1.6K
- Followers83
- Likes3.8K
Supposedly, 6.2 Astra is the model responsible for the math drop?
From my personal experience trying this model, I am dubious, but 🤷♂️
Replying to @davidiach
Yup it is the model that's good at math, and I'm looking forward to all the breakthroughs people find with it!
Corey retweeted
I've spent a good deal of the last two weeks working as an early tester of the new Astra model. This will be a longer post about my experience as a working scientist and mathematician, so stay put because this will be a wild ride.
First thing - I always use Astra in agentic mode (my preferred version is CLI mode in my terminal). I like seeing what is going on with the model (all the gory details), keeping control and seeing all the instructions which are executed. New Astra is a very perceptive model and keeps doing things incredibly fast and efficiently. Keep that in mind. No YOLO mode. Always stay a bit alert.
Capabilities: for mathematics, especially mathematics driven by hardcore computations, this is a killer model. I usually start with a very specific auto-research task, driven by the labyrinth skill which I developed, trying to figure out the constraints for what makes the particular theorem possible and why. This model is a heavy weightlifter. It requires patience. It is very common that once the setup, preps and first few runs are done, the model goes on a research hunt for a FEW HOURS with a simple "Carry on". Yes, most of my runs in the cycle were about 1–4 hours each. This is a lot of reasoning, a lot of grinding and a lot of new ideas coming through.
Partner: this is a model which I use both in conversation and in agentic loops. It can discuss very fine details of research with you and never get derailed or distracted. Doing things is very natural once you master the work mode. You should focus with this model (I prefer Max over Ultra) on a particular task. It's capable of going very deep. In my work on hypergeometric motives (all minimal polynomials of finite ones), the model proposed a very mature strategy. That was the work of a very capable mathematician.
Speed: when you turn on fast mode or lower the effort, you get a smart agent which is amazing for day jobs, but the most powerful version of Astra 6.2 comes from working in the Max tier and then it gets a bit slow. It processes a ton of data and its own chain of thought. But the output mostly works in one-shots.
Writing: the writing skills got much better with this model. I finally like reading the production drafts which the model produces. It works best with a specific prompt, but it's much, much better than Astra 6.
Agentic: this is what I got the most excited about. I was classifying semigroups of order 7 (almost a million objects), rewrote Mizar into a fully modern kernel with modules and tactics, and got a breakthrough in my research on elliptic divisibility sequences. All of this worked because the agentic orchestration and division of work were relatively natural. You need to work hard on this part yourself - the vision of what you want to achieve. The model won't eagerly set up a ton of agents for you, but you can trigger it to do so and even make communication channels between them. Powerful thing.
Lean: this is where I find this Astra really cool for maths. Once you have a ton of AI slop with great ideas, doing a Lean formalization attempt is a piece of cake. If it's harder than expected, Astra will point you to the hard parts and let you focus on checking the theorem formalizations. For my semigroups project, it produced a few million lines of reusable and very structured Lean code in two weeks. With AI-generated maths, I think Lean is a necessity.
Overall, this premiere comes with no bells and whistles, but it's a powerful tool for power users. Sol is a tiny bit less capable but is a much faster and cheaper agent. It's my go-to tool for everyday work. Both combined make a totally awesome tandem.
Replying to @OpenAI
GPT-6.2 Astra—a frontier model built for your hardest projects. It sets new highs on evals across professional work and scientific research at less than a third of the cost per task of GPT‑6 Astra.
You’ll see the difference in things like slideshows, websites, and 3D designs.
Corey retweeted
I’d note that OpenAI has always been great about not influencing early testers & has encouraged candid feedback. The mixed initial 6.2 reactions from testers didn’t blindside them.
Some model point releases are just weirder, I wouldn’t overindex on it as a sign of the future.
Corey retweeted
I think this interview between @sacca and @Alfredijohns0n clearly captures how a lot of people in clean energy feel about enviro groups right now over the permitting bill:
“it's exhausting because the reality is those of us who are actually trying to build solutions at scale, we find the people we run into most as the obstacles to getting stuff done are the activists.
it's actually more the well-funded environmentalists who just sue at every point, who take advantage of NEPA and CEQA and everything else to just stop actual progress in the space. It is frustrating because we should be on the same team right now. But that's the biggest movement we're up against right now: the obstructionists.”
A great interview: crux.com/critical-capital/in…
It’s genuinely crazy how Democrats write memoirs like this. Biden doesn’t even try make himself look good here. For the history books I suppose!
And this is exactly what the reporting has said occurred during this exchange for years, too 💀
28
Corey retweeted
Ha. When I had Astra 6.2 make a puzzle spy game using any setting it wanted, it set the game in "Bellwether Station." I thought it was an odd choice but didn't realize how deep a pattern this was. What a weird attractor for the model.
I was saving this for my video but I'll spoil it now.
Ben and I both made new levels in games using 6.2 Astra during the early testing. He did Majora's Mask, I did Outer Wilds.
We both prompted ourselves, but it was along the lines of "make a creative new level for [game]"
We both got water levels with music themes, specifically around bells.
Both were named "Bellwether".
This model regresses to the mean like nothing I've ever seen.
Go ask it to make a new level for a game you love, lmk if it chooses "bellwether" as the name.
Corey retweeted
One criticism of various modeling efforts on climate impact of #BAAJA is that it did not consider emissions from increased #LNG exports.
We ask the counterfactual: what happens to global emissions if US LNG exports were banned?
Read on! 1/
Paper: chemrxiv.org/doi/full/10.264…
Corey retweeted
One of the funny reasons for the difference in vibes between the influencers and the crowd today on 6.2 Sol/Astra is that we gave a few folks access a while ago and then made a lot of changes in the last 3-4 days and I'm pretty sure no re-testing was done within those days and the videos were produced last week.
Don't get me wrong, while the models are a hefty jump compared to their GPT predecessors, they are by no means perfect and there are a few things the teams are excited for the models to get better at in future iterations in terms of taste and niceness of use. Love all the feedback and we hear you!
Corey retweeted
A relaxation of zoning restrictions in Raleigh, North Carolina, led to a surge in construction, which reduced inflation-adjusted rents by 15% between 2022 and 2025.
Low-income renters benefited the most.
pew.org/en/research-and-anal…
It’s crazy how grad students in these field are too stupid to realize “90% of new US energy in 2025 was renewable” even before BAAJA, mechanically disproves that BAAJA is a “methane bomb” if you believe it speeds everything up, even data centers.
The climate case for BAAJA assumes it benefits all energy sectors equally. Data centers don’t work like that.
90% of new US energy in 2025 was renewable. Yet due to siting, only 15% of permitted data centers are on renewable.
This bill is a methane bomb.
1
43
Corey retweeted
1/ Public intellectual life is absolutely *impossible* if people in positions of epistemic authority -- and this is a politics reporter at Rolling Stone -- are unable to complete basic, high-school level reading and listening comprehension tasks.
I have long been obsessed with Holocaust and genocide curriculums because more often than not they fail to impart on students an understanding of the crime and its sociopolitical causes and instead leave them with a very rigid, industrial understanding of genocide. One that holds the Holocaust as the textbook example, and everything else as a partial/ lesser example.
Really it's the opposite. The Holocaust is exceptional among genocides because of the level of industrial resources used in the commitment of the crime, and the academic and cultural weight it produced.
Most genocides don't look like the Holocaust post-1941. It's not usually gas chambers and trains, its guns, machetes, starvation, and disease. The Nazis were not Nazis because "they had gas chambers." Before the gas-based extermination camps came into operation in 1941, the Holocaust was carried out with guns and mass burnings in Eastern Europe. Millions were killed in ways that mirror other genocides.
Ezra uses the gas chambers here as a an attempt to undermine the notion that Israel is committing a genocide, by arguing that it doesn't LOOK like what most people were taught a genocide looks like. It’s not a definitional argument, or a philosophical one, it’s purely material.
It's insidious. Klein is not dumb. He knows most of his audience is not familiar with Article 6 of the Rome Statute. The people listening to this probably received a standard american Holocaust education im which they spent a day in the WWII unit discussing the liberation. They maybe read Night and Anne Frank, they maybe watched Schindler’s List or Life is Beautiful.
How to make it all more effective, and useful to understanding current events, is something Holocaust and Genocide educators continue to discuss and debate. How do you impart a broad understanding of the crime while also preserving the important context and origins of the academics?
Corey retweeted
NEWS: Wall Street execs and NSA have told Congress that outages were caused by patching vulnerabilities identified by AI models, sources say
AI did not cause outages but found so many issues that the rush to fix them before an attack overwhelmed systems
washingtonsun.com/technology…
It seems to me it was more about how we could use a technology that was becoming rapidly cheaper to not only lower the cost of living, but also mitigate climate change. Win win.
My guess is that the centrist focus on climate change in the 2010s was basically just normal group psychology. A thing to worry about to sound serious and adult. Doubt many were actually personally that anxious.
44
Corey retweeted
Thought more about this. Anthropic’s lead started with Fable 5. It was the first true “next gen” model.
I honestly think most of yall missed just how far ahead Fable was at launch. Understandable tbh. It required you to use the model entirely differently. It was prohibitively expensive. It got taken from us after just 3 days of usage. It had no ZDR policy, so you couldn’t even use it at work in most places 🙃
The result: most people perceived the gap between Fable 5 and GPT-5.6 Sol to be small. IMO, it was massive.
GPT-6 Astra mostly caught up on raw capability for short term tasks, but it was unable to stay coherent for long runs.
Meanwhile, Opus and Fable keep getting easier to work with and more coherent on long runs, while also gaining in capability. Opus 5.5 brought the Fable wins to everyone, by making it way cheaper, more generous on sub limits, and also great for enterprise (zdr!)
OpenAI has maintained pace with capability, at the cost of coherency and readability.
Tl;dr - Anthropic has been ahead since Fable 5. OpenAI’s attempts to catch up on benches have hurt the model pretty much everywhere else
Corey retweeted
Both 6.2 Astra and Sol are regressions on Artificial Analysis’ Coding Agent Index versus their predecessors, lagging all recent Claude models
Yikes 😭
Well…that’s exactly the point. Monopolies *are* rational for the monopolists.
If there’s no competition to deliver cheaper power, then yes why would a utility make an investment to do so, like more transmission that combined with renewables would likely lower costs?
You should think critically for just a second here.
If an utility is well hedged with generation and they have sufficient contracts to ensure their resource adequacy WHY ON EARTH would they want to pay for a share of transmission that doesn't guarantee capacity that enables "low cost" generation that is redundant with their hedges that they and their ratepayers are on the hook to pay for anyways.
This angle of "evil utilities blocking cheap energy" is asinine. It often doesn't make sense for exactly this reason. And forcing LSEs to pay for transmission so that some corporate can let their $50 PPA intermittent energy displacw $50 hedged energy is not exactly the most economically beneficial thing!
This is the reason the munis and coops fought so hard to be exempted from this smdh!
Corey retweeted
This just isn't true. Per @rff modeling, electricity costs decline immediately if we pass #BAAJA, in part due to interconnection queue and project certainty reforms getting more cheap clean power on the grid fast.
Senators should get this stuff right.
rff.org/publications/issue-b…
Louder for the Senators in the back, @SenMarkey! Vote NO on BAAJA!
"The permitting bill sells out those communities for the promise of electricity bills that might go down by between $4 to $6 per month in 2040. Can I say that again? Your bills don’t go down until 2040! And it’s only $4 per month. Is that really worth it?
We have Pages out here that are 16 years old. They'll be 30 years old before they see the benefit of this bill! That's not what this election was about."





































