Benoit Dubosson

Imagine a breakthrough in inference that’s model-agnostic and infra-agnostic across GPUs and ASICs Same workloads No loss in performance Roughly half the cost At 2x the speed I guess it’s a fantasy but just imagine if it wasn’t! How cool would that be huh

Need to get this off my chest I envy those who genuinely don’t understand the concept of assisted suicide and comment things like “it’s never the solution” They have so obviously never gone through, or even just seen, with their own two eyes, what agonizing pain does to a human being There is a point of no return once you do, almost like a switch and after that almost everything becomes understandable I know it’s not socially acceptable to say that suicide is understandable in certain circumstances but regardless of what your moral compass/religion says, it is… And if you disagree you can light a candle tonight and thank the universe for the life it gave you and your loved ones You can be morally against it, I understand that in some cases it goes against your religion and, I will fight for you to have the right to refuse to use it, the same way I fight for your right to refuse organ donations even though I am a donor myself All that I ask is for you to allow others to have the right to make that decision for themselves You think they will go to hell? I understand, but they made that decision, just like the other millions of sinners out there who made theirs I am gonna be very grim from here on out but it’s to make sure my point gets across When I was 8 I was in the intensive care unit with another kid who had lost his battle with leukemia, he had maybe a couple of weeks left max and more importantly he had an extremely severe case of mucositis from chemo Essentially the radiation from chemo killed all his mucous membranes (throat, mouth, stomach etc) Basically imagine you have open wounds from your mouth to your stomach, it feels like you’re burning alive every time you take a breath He screamed days and night to the point that he ruptured his vocal cords, parents kept refusing to put him in a coma knowing he likely wouldn’t have the strength to wake up again It had gotten so bad that its the only time I recall getting moved out from the intensive care unit because nurses were, and rightfully so lol, scared for the impact on my mental health I don’t know what hell looks like in your religion, but what I saw/heard that day cannot be significantly worse than how it is over there I still vividly remember him screaming or more so screeching from the top of his lungs “I am burning please let me die mom I beg you let me die” Over and over until he lost his voice and was so sedated he was barely alive until he passed a few days later That night, I was 7 years old and yet I remember it like it was yesterday I put both my knees on the floor and my hands together on my bed, mind you I was not religious, and I prayed saying “God if what they say is true, if you really love everyone and so you love him too, please kill him, and if you cannot do it yourself because you’re god send me a gun and I will put him out of his misery myself” Besides the fact that I didn’t quite understand how religion or praying worked, I understood pointless suffering aka as I call it: “Legally imposed torture” All hope is long lost by that point, death is not only inevitable it is knocking and the only thing that predates it is pointless suffering Go speak to one of these patients, look them in the eyes as they’re filled with nothing but despair and I guarantee you you that you’ll never see the world the same way again The deepest humanity is found in easing another’s suffering, even if at times it means saying goodbye

@martin_casado @drydenwtbrown Thank you

Because nothing in the data supports it. The big labs have been able to aggregate more money as a percentage of total equity capital than any other time in the history of the industry. And the same goes for pricing power (outside of a fairly direct price war between the large well funded labs). All this while open source has been readily available. Further, third party margins are quite strong and when you see erosion it's exactly due to the subsidization from aggregated capital of the large labs, not open source. I'm not feigning incomprehension. I pointing out that it is simply wrong.

It’s a simple argument. Open-weight models compress frontier model margins, which reduces cash to invest in scaling up compute for larger pretrains — which is where material increases in intelligence come from. Why feign incomprehension?

@AgustinLebron3 @aakashsabharwal Lol

@aakashsabharwal The tranche two-step scam.

How does one make sense of a company having two valuations at the same time?

@deanwball I am actually amazed that you had the audacity to post this I genuinely refuse to believe that you are sincerely sold on your own arguments

Some observations on Kimi: 1. It's a very good model! I don't think its performance can be explained away by distillation or anything like that. In agentic coding sessions, it seems pretty much on par with the best public models of Q1 2026. In my fairly limited use, it also seemed very token hungry. It's not obvious to me that this model is actually that cheap to run. 2. I am personally surprised the Chinese state continues to allow the open sourcing of models this good, given potential risks. To be clear, I *myself* might be fine with models presenting this level of marginal risk being open weight, but I am surprised that China is fine with it. I suspect the reason they are is 75% explained by strategic blindness/lack of AGI-pilledness (the CCP is very Yann Lecun-y in its views of AI). The other 25% or so is their lack of compute for customer inference (making China's open-weight strategy an unintended byproduct of US export controls) and the normal Chinese strategy of aggressive exports. For the companies, as opposed to the government, the decision to open source is partially ideological and partially because they are behind, and they know that very few people would pay for sub-frontier models from China. 3. Open-weight models are inherently decelerationist, and I'm continually surprised to see the so-called "accelerationists" so excited about open-weight models. I suspect the reason they are is that they know open-weight models are effectively ungovernable, and they simply like the overall cloak of ungovernability open-weight models create over the whole of AI. It's not a bad strategy; it reminds me of James Scott's recounting of the hill people in "the art of not being governed." Still, in the end, open-weight models deter further AI capex. 4. One probable outcome of an open-weight-model-dominant world is full AI communism, which is precisely what China proposes: rather than a market product, AI is a "public good" which will ultimately be provided by the state as a kind of "digital public infrastructure." This future strikes me as a dystopian hellscape, but I've never met an open-weight models advocate who doesn't ultimately concede this is where things end. You'd be surprised how many 'accelerationists' lobbied me, while I was in government, to support an eleven or twelve-figure federally funded data center so that startups could train models at a subsidy and then give them away for free. There was no other way for AI to progress, they said. Perhaps this is the logical end state of things. Nonetheless, I find myself surprised to see supposed accelerationists excited about such an outcome. I think many of them just don't know what they're doing. Many accelerationists do not view the creation and serving of frontier models as a legitimate business. 5. I would guess that the Trump Administration will at some point realize that their best strategy here would be to create large amounts of regulatory risk around the use of open-weight Chinese models. You don't need to "ban open source" (one of the dumber motifs of AI policy discussion). You just need to direct every agency to issue soft law that creates FUD. "A Federal Reserve Advisory Bulletin found that there may be backdoors in Chinese AI models." It needn't be that well justified. You just create enough regulatory risk that every regulated enterprise backs off. You probably don't want to create so much regulatory risk that you scare off the hyperscalers from serving Chinese models; this will just drive startups to sketchier providers. There's a happy middle ground here. I'd assume they will do some version of this. 6. It's probably true that open-weight models of this capability make the world a bit more dangerous, but not so much more that you'll really notice. At some point the models will be capable enough that you will notice. "A nonliving, invisible, dangerous, and infinitely self-replicating agent escaped from a Chinese lab," you say? Color me shocked.

@msuiche @MistralAI Guess we all did

K3 is what I was hoping @MistralAI would do.

@zeroxkyle The argument is horrible as well lol

can someone explain why the head of strategic futures at OpenAI thinks that AI being a public good is a dystopian hellscape?

@ScottPh77711570 @cryptohound666 @TheNestintern @Cbb0fe Wait wdym Scott I think I am confused? You would take the naked position on HL without hedging on IBKR? Do you then just throw some shit on to hedge beta or what? I used to just pay for the hedge so I could sleep at night

@cryptohound666 @TheNestintern @Cbb0fe If the good leg is HL (it should be obviously) then it is highly likely that the HL account has all of the alpha and a lot more variance I would (not copytrade but I would trade this on hl only)

. @Cbb0fe has been number 1 equities trader by uPNL on hyperliquid for the last 24h and 7 days started copytrading him today for new trades, will be sharing results Hyperliquid

@clanker_ Fable I think is 6+T with 200-500B active parameters Maybe @sasuke___420 knows, he’s much more knowledgeable than me on such things

@clanker_ Opus 4.8 is <2.5T But with 100-150B active parameters GPT 5.6 not sure tbh I expect something like 4T with also 150-200 active parameters

Fable is probably ~5T params & undertrained if i had to guess

@liquiditygoblin @disconcision @sasuke___420 I’ll make one for you boo

@disconcision @sasuke___420 @beniduboss can you guys make some of these as customer gift/manuals?

Switzerland was ahead of the game by open-sourcing *checks notes* their entire government

@0xLouisT Très très fort

As a Parisian, I can confirm this is TOP TIER ragebait: 99% of Americans won't get the joke

@quant_arb @DrewVento Sell it to Berkshire and all the companies supplying liq to insurance providers

@DrewVento Why would PE buy it. I think you’d be better speaking to JS

I’d sell my option to press the yellow button to a PE firm for $490,000 dollars

man ignore the benchmarks and the race i just want there to always remain a frontier level intelligence model that is open and safe, available to everyone regardless of region it is truly incredible to have built such an open (and strong) model with a group of genuine people

@blknoiz06 Can’t wait for the next wave of influencers to just be guys who hit >+3σ variance and accidentally became multimillionaires

if i could long ANY kind of media, it would be this: the next celebrities are the ones turning $1K into MILLIONS in public, documenting every trade, then teaching you how they did it these will be the most popular people on the planet. i promise you

@Icebergy Actually good Will test properly next week

is Kimi K3 that good? or are the ai influencers getting paid to post how great some newest model is again?

Kimi Kimi Kimi a man after midnight Won't somebody help me chase the shadows away? Kimi Kimi Kimi a man after midnight Take me through the darkness to the break of the day

@Rager People still don’t know that paper wallets exist

Most hardware wallets are garbage and I've almost lost crypto multiple times due to hardware and software issues with wallets You're more likely to lose your Bitcoin and crypto in a hardware wallet than you are keeping it on a separate phone or even sitting on an exchange like Coinbase or Kraken

@Versace_Trader @_RogueTrader @zoltar02 That was the only other option so makes sense

@beniduboss @_RogueTrader @zoltar02 I’m almost certain he’s Indonesian. He seemed to be very upset about some election stuff