MMLU-Pro:
Gemini-3.1-Pro at 91.0
Opus-4.6 at 89.1
GPT-5.4, Kimi2.6, and DS-V4-Pro tied at 87.5
Pretty impressive
HN user
MMLU-Pro:
Gemini-3.1-Pro at 91.0
Opus-4.6 at 89.1
GPT-5.4, Kimi2.6, and DS-V4-Pro tied at 87.5
Pretty impressive
Gotta derail any peace talks!
I can have an LLM generate a 10,000,000 article encyclopedia and be better than both!
Worth noting that a partner at Sequoia (Shaun Maguire) publicly accused the wrong guy of being the shooter.
https://www.fastcompany.com/91463942/sequoia-shaun-maguire-b...
Claude summary:
Key Findings:
Trump-Epstein relationship:
Epstein frequently discussed Trump in emails, calling him "the worst person" he'd ever met with "not one decent cell in his body" Despite this, evidence shows ongoing contact, including Epstein visiting Trump Tower a week after the 2016 election Epstein's pilot tracked Trump's movements to Palm Beach Epstein claimed to know "how dirty Donald is" and called him "borderline insane"
Continued elite associations:
Despite Epstein's 2008 sex offense conviction, prominent figures maintained contact:
Peter Thiel: Discussed meetings with Woody Allen and dinners together
Steve Bannon: Extensive correspondence about European politics, think tanks, and strategy sessions
Ehud Barak: Collaborated on Syria-related op-eds
Troubling journalist relationships:
Landon Thomas Jr. (NY Times): Advised Epstein on PR strategy and requested "photos of donald and girls in bikinis"
Michael Wolff: Consulted Epstein while writing books about Trump, sought his input on framing him as credible
They have territorial disputes with their neighbors, that’s not the same as trying to project their cultural values globally.
What dispute do they have with Pakistan? You mean India?
Feels unnecessarily jingoistic, China succeeding will not come at our expense. Unlike the US, they don’t want to project their values across the globe.
Who’s working on gene editing for IQ maxxing?
Larry Ellison getting TikTok from the govt at just $14B is the biggest steal. His control of Paramount and CBS makes him the biggest owner of American attention, and he’ll have Bari Weiss steering it.
I like the idea! In my opinion (looking at SF) it’s still too low resolution. SF in particular can vary greatly in safety, walkability, etc. even a few blocks over within a neighborhood.
Because 1) this is not their own citizens they are surveilling and 2) Israel is committing genocide according to many (most?) international and human rights orgs, which would make Microsoft an accomplice
“Jordan this week said they were told that Israel had prohibited filming Gaza from above.” Prohibiting reporting of a war zone should be seen as automatic confession of war crimes
The US continues to burn its soft power capital to defend Israel - this guy was only targeted because of his investigations on Israeli war crimes.
What has China done that comes even close to what the US has done or supported all over the Middle East, esp Iraq and Gaza?
It seems like Elon also still is a fan of JD (he suggested that Trump be impeached and JD become president).
The Palantir infiltration of govt and surveillance will continue.
The speed of the implosion is still surprising, as is the fact that Elon mentioned Epstein. It is a much bigger deal to say that Trump is a pedophile (under blackmail by Mossad/CIA)
Was there any trade off using AI? Like limit in customizability due to using prompts that would not be the case if you hired someone.
Output video looks very cool
Palantir has been playing a large role in these deportations, using AI to detect “antisemitism” (opposition to Israel) and handing student lists over the Trump administration. Paul Graham recently attacked them over this.
The problem is that people are not feeling inspired by the future, where greatness is inhuman AI. Interesting argument coming from Noah, who is a big fan of the American empire and every war it involves itself in (Ukraine, Israel) and constantly attacks China and Russia. Claiming China is in a state of steep decline is a wild statement.
Gradio is built with svelte https://github.com/gradio-app/gradio
Worth noting the Sheryl Sandberg produced “Screams before Silence” which featured many debunked claims about Oct 7.
The Law Review site had an article that cast Israel negatively, so the board of Columbia Law had it shut down.
https://apnews.com/article/columbia-law-review-israel-articl...
The US and even Israeli governments consider these numbers as accurate. The Health Ministry provides the social ID numbers of every death so the deaths would be impossible to fake.
Almost certainly this is some prompt post processing, where the prompt is appended with “Make the image diverse in race and gender”. This is a trivial fix and not baked into the weights. However removing this would almost certainly surface the actual overrepresentation of “men as lawyers, women as nurses” etc. in the training data and thus the weights. Google tried a simple hack to remove this bias via prompt engineering. It’s stupid but overplayed
For those missing context, Paul Biggar released a blog (https://blog.paulbiggar.com/i-cant-sleep/) decrying what he saw as a massacre of Palestinians conducted by Israel, and supported by what he considered the overwhelmingly pro-zionist tech VC industry. He called out some of his own investors by name.
Paul Graham has tweeted similarly, and also called out collusion in the tech/VC industry to silence Palestinian voices.
Cancel culture in tech. Even YC (Garry Tan) cancelled him.
It correctly answered `The name of the cat that wears a blue tie is Jackson.` for me
Most real world work is grinding through boring work as well, and it’s worthwhile to build the discipline for it.
A company is more than the personality of its CEO. Gates, Zuckerburg or Elon's personal faults shouldn't override the work and mission of thousands others.
I doubt it, that would kill their whole brand.
It's a clickbait / not informative title.