Anish Acharya @illscience

@VCBrags @KennyisGaming Oh, cool. Definitely interested in your view @KennyisGaming

Gaming has got to be the market with the most shattered dreams, unkept promises, and beautiful ideas that never saw the light of day. People get into it for the love of the craft and end up resentful, ground down by the realities of production schedules and brutal market forces. If nothing else, it’s going to be awesome that anyone with a game idea can finally see their creation come to life. Even if it’s just to discover that their idea isn’t actually that interesting, and they’d rather work on someone else’s better game instead.

@joshelman Me too!!

what we're picturing vs what we're getting 😂

Tim Sweeney @TimSweeneyEpic

a day ago

The games we've long wanted, which didn't exist because they'd cost several times more to make than they'd earn back, will soon become possible. As will the games developers wanted to build, but couldn't with the resources available. It'll be a new golden age of gaming.

2K 1K 16K 5.2M 2K

Sooo incredible. I feel like we’re reliving the early Internet all over again. But juiced 10000x

Dmitrii Eeeekimov @Ekimko

2 days ago

Лучшее, что я видел за последнее время — чел пилит ASCII MMORPG, в которой вся графика состоит из букв и символов. Да, буквально весь мир, персонажи, окружение и спецэффекты нарисованы разноцветным текстом. Автор сейчас экспериментирует с магией, визуалом и боссами. Я прям

keyboard_arrow_left Previous keyboard_arrow_right Next

322 2K 18K 550K 7K

today we open the beta for Krea 3, a new frontier for image editing. - end-to-end 2k edits in under 8s. - matches Nano Banana Pro or GPT Image 2.5 quality. - bounding box control and precise multi-turn edits. learn and get access 👇

I agree with the overall point, Lenny. For some reason, there is a moral panic about moats, even though most of the classics are unaffected by large models. HOWEVER - I believe most switching costs are going to zero - since prohibitive migration complexity for humans is trivial for agents. Coding agents are really really good at systems integration and migrations. Even though switching costs are usually discussed in an enterprise context, the single biggest profit pool that's protected by swictching is consumer financial services, which are often mispriced to the consumer's detriment. That's why JPM has blocked personal agents, but that's not a long-term strategy. Decreasing switching costs is the real saaspocalypse to come IMO - everything from ERP to payroll systems to consumer deposits should move more fluidly and rationally, and this great reset is going to really shake up the snow globe of a lot of companies' profit pools.

@ericosiu Thanks brother 💪

@zkevinbai 💪💪

Re everyone can win — yes. Positive-sum outcomes are as hard to intuit as exponential growth, so we tend to underprice the possibility. IMO we’re in the greatest positive-sum melt-up of all time, where we get to break every trade-off since the Industrial Age started. GDP can grow 10% AND the workweek can compress to four days AND the qualitative experience of every white-collar professional and individual consumer can go way up. There are lots of precedents for everything getting better — no one would argue that life was better in the Middle Ages, and life today may seem similarly brutish in retrospect. To use the Baptists and Bootleggers frame, there are a few earnest Baptists who are extrapolating from progress in areas like math/code to non-verifiable domains in ways that probably don’t make sense. Further, Dario has talked about economic diffusion as a point of attenuation for model impact, and anyone who uses models at work can tell you this is true. We may also simply be overestimating the number of problems that are intelligence-bound. Despite the warnings from Bootleggers whose business models are threatened by a frontier monopoly on tokens, the reality is that token distributions in the enterprise should be very heterogeneous just as different job functions select for different human traits. The best finance leaders are neurotic, and the best design leaders are highly open. It’s unusual to have a model that can capture outlier performance in these two opposing directions. In my view, the most rational model that matches what we’re seeing empirically is a hybrid of frontier, open, and emerging models like Jev that represent a totally different point on the Pareto frontier. Finally, I think that we're pushing what's possible at a species level to a degree that we haven't ever seen before. It's everything amazing that's happening thanks to SpaceX. It's the rapid possible progress we may make in disease and longevity, and all of the theoretical breakthroughs that were on the doorstep thanks to agent swarms and models. No one tells you when you're in the good old days, so let this be your warning :) All of which is to say, Gavin is brilliant and right. I think it’s all going to work.

Gavin Baker @GavinSBaker

5 days ago

1) Trying again. 
TLDR: Everyone can win. Open-source, frontier labs, infrastructure, applications and most importantly the American people.   I understand all of the jitters about Anthropic losing share, real rates, spreads and second derivatives. Still think the most likely

185 394 4K 837K 6K

@OfficialLoganK Congratulations!!

@rupespupkin 🤣

Yes, I used "melt up" a little loosely here, but it was more fun to write than "boom."

@finkd @giffmana Fable class model from Meta?! Wow..

@zkevinbai Beautiful!!!!

@avithaldas @typesafeai Yeah, but over and over and again we’ve seen that for every category of model there’s comparative advantage / specialization in different directions, and the market is just way, way bigger and more expansive than it seems at first blush

Congrats @typesafeai !! What's so cool about Jev is that it breaks the tension between expensive, dynamic intelligence at runtime (w/ opex costs) and cheap, brittle intelligence at compile time (w/ capex costs). More: - One tension between labs and their customers is the economics of coding agents vs general agents w skills / projects / memory - We went from fat apps (most of 2025) to thin apps that mostly wrap intelligence, with code as a disposable artifact generated at runtime - Generating code is a crystallization of inference. It captures business logic that could otherwise live in the model as static procedural logic. Cheap to run, but brittle - The alternative is regenerating everything dynamically at runtime with PhD-level models w reasoning, memory, tools. More elegant / general, but the most token consumptive: you pay opex every time you use the app vs capex once - This partially explains why labs have pushed so hard for skills / projects / agents over code first - the general agent approach drives far more token consumption - Most runtime "intelligence" is small decisions: classify / route / score / pick. A fast, cheap System One model gives you dynamic intelligence at runtime for roughly the cost of compile-time code - So the future stack should be code generated by models at compile time, Jev-style decisions at runtime, and frontier agents + tools reserved for genuinely open-ended work I LOVE this model & I'm very excited to see what comes next.

Jennifer Li @JenniferHli

a week ago

x.com/i/article/2108…

29 20 321 145K 136

@dpetrou @typesafeai Totally - I said every morning for email triage, and it’s blazing fast and very cheap.

@dtrinh @boneGPT @Muse 😂 incredible

What’s so interesting about this is that SpaceX is getting to eat their cake and have it too - they get the benefits of being a lab, vertical integration into inference, and being a Cursor style router. They have a SOTA model so labs can’t bully them, they can capture margin on 3P models through inference, and somehow (Elon!) they have access to Suno and MJ as well. They can offer the overall best, the pareto best or the best for a tail modality and the house always wins. Amazing on so many levels.

Elon Musk @elonmusk

2 weeks ago

Important note regarding Grok @bot: Going forward, @SpaceX will use the best back end model for any given task, including Claude Opus 5.5, MidJourney, Suno and other leading APIs. Whatever is most likely to give you the best outcome.

7K 8K 108K 79.7M 10K