Altman won't go public for less than $1 trillion, so OpenAI's IPO may slip to 2027
Advisors are telling OpenAI to hold off on going public until next year. The triggers: volatile tech markets and SpaceX's weak stock performance after its record…
Advisors are telling OpenAI to hold off on going public until next year. The triggers: volatile tech markets and SpaceX's weak stock performance after its record…
RT SpaceXFull duration single-engine static fire test of Starship
RT DogeDesignerBREAKING: Venezuela’s Vice President Delcy Rodríguez thanks @elonmusk and @Starlink for providing free connectivity after the earthquakes.
RT Victor Ma bit more than just another AI company
TiresomeBAD👏THINGS👏CAN👏BE👏BAD👏EVEN👏IF👏THEY👏DO👏NOT👏HELP👏CHYNA👏WIN!👏"Hurting China" is not a reasonable terminal value! If you turn your society into China-damage-maximizers, you'll also have a bad time! Yes even if this isn't…
Article URL: https://github.com/mondaycom/HATCHA Comments URL: https://news.ycombinator.com/item?id=48685677 Points: 42 # Comments: 44
RT Gunnlaugur JónssonRe When you came to Iceland in early June 2025 we had not yet started to see many of the consequences of the 2024…
Consider that after the Russian Revolution such men joined Cheka en masse, and much becomes understandable about the 20th century historyScience Banana: the montreal shooter's manifesto…
RT Yvonne KussmannHeute vor fünf Jahren, am 26. Juni 2021 fand man in Wien am frühen Morgen ein totes Mädchen auf der Straße. Leonie war gerade…
As agentic systems continue to evolve and are widely deployed in real-world scenarios, there is a growing demand to faithfully evaluate their capabilities. However, current benchmarks…
Multi-model LLM systems such as routing, voting, cascades, fusion, and mixture-of-agents are used to beat single-model accuracy. We show that their gain is capped by a…
Article URL: https://alephneuro.com/blog/ultrasound-brain Comments URL: https://news.ycombinator.com/item?id=48685558 Points: 25 # Comments: 7
RT Ziad ObermeyerMy friend the brilliant @kevin_volpp dropped deadHe survived (read abt it: https://tinyurl.com/ypcpdzpz)But every year 300-400,000 in the US don'tThis week: our @nature article (+A.Schubert…
CUDA: add cublasSgemmBatched mapping for HIP/MUSA vendor headers (#25033) macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED macOS Intel (x64) iOS XCFramework…
> these are American tokens! our tokens! you stoled them!I like the "token stealing" meme, hope to see more of itOutdated Often: @askalphaxiv Ok and so…
Russia freezes roughly a third of its datacenter constructiontwo reasons given: high key rate, and impossibility to contract for new power capacity (!!!)(the third may be…
RT Victor M300tok/s on mobile is insane... open source must win ✊Xenova: While we eagerly await Fable 5's return, our agentic WebGPU kernel optimization framework kept…
"Returns on intuition": Why Anthropic no longer needs junior engineers and warns of an economic shock. The article Anthropic doesn't need junior engineers anymore thanks to…
Not even ZAI claims they're close to Fable/Mythos (they are on some very narrow benchmarks, but they're not so petty as to inflate this). Cade is…
Nixon today would not even get elected in the US. The premium on theatrics and lowbrow pandering has risen colossally. But on the substance – I…
RT Sergio PaniegoTRL v1.7.0 is out‼️+ continuous batching makes GRPO and RLOO 1.25x faster at -16 GB+ proper MoE post-training across GRPO/RLOO/AsyncGRPO+ new GMPO trainer+ AsyncGRPO…
Spent two weeks running 40 coding prompts across 7 models, self evaluating each output on correctness, completeness, and whether I'd actually use it without editing(which to…
This really is somewhat insane. Meme numbers. The problem of course isn't the balance of trade with China, it's – where do Europeans get the surplus…
So, the US is adding… just 12.3 GW of power/year?That's good, actually. China adds only something like 47, in pure 100% capacity factor terms.Of course, none…