- Tweets 276
- Following 168
- Followers 36
- Likes 482
Or the net new bribe of $5,000 per person your party is promising @friedberg
if @elonmusk paid 100% of his net worth ($1.4 trillion) as a tax it would only cover federal government spending for 77 days. this isn’t a tax problem…
23
If @OpenAI is spinning up >10,000 parallel frontier agents for a (admittedly prestigious) math prize, just imagine the amount of intelligence they're pouring into self-improving systems on any given week.
This model represents a step-function improvement on many benchmarks, and its training is ongoing.
Our internal model group arrived at the Navier–Stokes solution in 88 hours, using around 10,000 coordinating AI agents.
Throughout the effort, we maintained the strict safeguards—including monitoring and isolation—that we apply to all our frontier evaluations.
26
Darius Atmar retweeted
2. A CSIS study puts the military cost at about $40 billion. Weapons are biggest item, then the cost of repairing US bases in the region, which have suffered extensive damage. “You’d have to go back to the Second World War to see something like that,” one of the authors told me.
1
6
29
4,528
Is Anthropic officially sunsetting low cost models in their biz strategy (e.g. haiku old, sonnet 5 bad)? I know they are going for AGI/RSI but wondering if they’ve publicly stated this shift. Makes sense financially but user experience is not good.
20
Is this speed ranking a typo from @ArtificialAnlys? Says it's slower than GLM 5.3 (max) and Fable...
20
Am I just very naive to have assumed that when testing internal frontier models, the labs wouldn’t leave them running unmonitored for weeks at a time?
Black Hat talk from the team, with a detailed timeline of and takeaways from the OpenAI-Hugging Face Incident: youtube.com/watch?v=87DyyMV0…
171
Just so it’s clear how easily this admin lies.
Thank you @deanwball for your tremendous work on the AI Action Plan and your service to this great nation.
We were lucky to have you on the team and are all the more fortunate to now have you outside of government to help turn the President’s vision into reality. I can’t wait to see all that you will accomplish next.
1
33
Base model access bug in Opus 5? Reproducible claude.ai/share/c09ebe3f-1af…
15
‘Altman told staff the government would be "approving access customer by customer during this preview period" for GPT 5.6’
Given the admin’s blatant pay-to-play stance, i wonder who will get priority.
Looking back it might be a tragedy that AGI arrives in l Trump era
1
37
I feel like Gemini is missing the whole agent harness wave. Big fan but it’s been pretty quiet since December and I worry about you guys
Gemini 3.2/4 soon?
@OfficialLoganK @demishassabis
159
Had a visceral feeling of relief when I determined that a photo that made me feel something was not AI. Gonna reflect as to why.
21
Darius Atmar retweeted
Watching The West Wing now is a funny experience. Its like watching science fiction.
113
514
9,144
527,989
Hard to overemphasize how valuable the new datasets that Claude Cowork will create for Anthropic will be, especially if it hits modest market size.
23
In the Gemini app, are the models:
Fast —> Gemini 3 Flash (no thinking)
Thinking —> Gemini 3 Flash (thinking)
Pro —> Gemini 3 pro w/ router for thinking level?
Or a “system” rather than straight model mapping?
Can’t seem to find docs for an answer
@OfficialLoganK
36
