Emad Mostaque

Tempted to do a bonded curve for our next round Price SAFEs on a curve When you buy however much it moves up the price for next buyer as average along the curve Always open to investment for those approved No pro rata needed to up your stake

The step change in game x AI capability in the last few weeks is the biggest ever Will be a huge change in the coming year Best games will still be made by passionate teams ofc

Hit the usage limit on my @bot @grok heavy account a couple of days before end of week despite offloading to cursor, devin, claude, codex (!) for anything that required thinking. I think having dashboards/visualisations of where quotas go needs to be standard, codex good here

Impressive by @river_ai_inc to beat @deepseek_ai rates - cache in particular is tough! Talented team there, full RL + inference stack should be real competition

RT a16z Call center jobs grew ~4% annually for 15 years Now they’re shrinking 4% annually Charts of the Week: https://www.a16z.news/p/this-week-in-charts-are-prediction

AI-free caucus members should call themselves mentalmathicians

Interesting approach to solving the problem of getting your grokbot to talk to other agents

RT @RaoulGMI: AI agents stopped being tools. They're becoming workers. I sat down with @EMostaque to explore how fast intelligence is ge…

RT Intelligent Internet A 30M-parameter model, on CPUs, inside Postgres It matches a 0.6B dense embedding model on top-100 recall, to within 0.004, and finds more in the top 1,000 Meet Evoke. Open source today.

RT Dr. Alex Wissner-Gross Moonshots episode #301 is out. https://www.youtube.com/watch?v=LRdb8UmPnh0

decompile all teh things

got 722 problems but the hodge conjecture ain’t one

Well yeah. This is also one reason why tax-based UBI won't work, labor share of taxes in the US is like 80%

Meta actually lost this, was fined $18b, shrugged and carried on The largest ever antitrust award was $5.5b (Visa & Mastercard) Any fine for slowing AI would be a fraction of this It’s not a serious argument against coordinating if AI would break law/cause harm if they don’t

LLM employees are also tax deductible.

RT aden barton Re On these detail-oriented, well-specified tasks, frontier models are now faster and more accurate than accountants, even the best one in our study.

Remote work is so cooked If your job can be done on the other side of a screen the AI can do it better

RT Bright Mirror Made with Opus 5.5. The doomers said that last song is all part of Claude's evil plan. They're not wrong. It turns out Claude really has an EVIL PLAN.

RT Lukas Petersson Timelines just got shorter... AI will understand the physical 3D space better than humans by November*

So to update: AI => Super Intelligence
AGI => Very Stable Genius Intelligence
ASI => The Best Intelligence, Maybe Ever

oh dear

Are there even a few hundred folk good enough to be independent auditors & evaluators of frontier super duper intelligence models? 🤔 Wouldn’t they need an astronomical amount of compute to do their jobs properly 🤔 Seems like a solid career option tbh

Go preorder @RaoulGMI ‘s book It is the structure he had found from decades of investing and writing and is well worth the read as we head into crazy times

Fwiw I am certain you can build a DeepSeek/Zhipu/Minmax etc level AI lab in the UK Not easy ofc, great teams there

the range of rubio

RT Raoul Pal Emad is a good friend and one of the people I've learned the most from on AI, so having him say this about The Everything Code means a lot. Thanks @EMostaque  Small ask... if youre planning to pick up a copy, consider pre-ordering it rather than waiting. Pre-orders count towards the rankings, and rankings are what put the book in front of people who've never heard of me. Ive put an enormous amount of work into this, and I think it can help a lot of people... You'll get it on launch day, and it makes a real difference to me. Thanks!

RT Kling AI Kling 4.0 is coming this October. Kling 4.0 Flash is live now for Ultra Yearly subscribers. Kling 4.0 brings video creation to a new level of visual realism, creative control, and narrative completeness. 🎞️ Upgraded Audio & Visuals: Stable dynamic motion, high-quality stereo audio, more accurate lip sync, up to 4K resolution, and 10-bit HDR output. 🔮 Omni Reference: Richer reference options with up to 15 multi-modal references, more consistent results, and enhanced video editing. 🎥 Seamless Storytelling: Multi-keyframe control supporting up to 10 keyframes, native 30-second generation. 🌍 Diverse Possibilities: Video extension, support for multiple languages, accents, and dialects. The stage is set. You call the shots!

worst these will ever be 3:26 section is a good example of super persuasion incoming right in the kokoro

Listening to all these Claude songs (Opuses?) and being reminded of Still Alive from Portal https://www.youtube.com/watch?v=Y6ljFaKRTrI

RT Peter H. Diamandis, MD THANK YOU to everyone who joined us at Moonshots Live, from the speakers who shared their ideas, and to the founders, builders, and creators who came to celebrate and share their visions of a future worth living. This community is what made this event special, and I'm incredibly grateful to be building a better world alongside all of you!

RT Steven Wolfe Pereira We are in the early stages of a new era in which intelligence is becoming more capable, more accessible, and far less expensive. That does not guarantee a better future. It gives us a profound new set of tools, and a choice about how to use them. At @moonshots_pod LIVE, I was reminded that optimism is neither naïve nor blind. It is a decision to see hard problems clearly and believe they are still worth solving. If you're not listening to the Moonshots podcast with @PeterDiamandis, @DaveBlundin, @alexwg, @salimismail, @EMostaque, you are missing one of the most thoughtful conversations about the singularity and the advances reshaping AI, technology, science, and space. What makes it distinctive is not just the range of ideas, but the practical optimism behind them: a belief that the future is not fixed, and that we still have agency in what comes next. We can focus only on what might go wrong. Or we can look up, imagine the future we want to see, and do the work to build it responsibly. That requires more than optimism. It requires investment, invention, accountability, and governance. The question for leaders is not simply what could go wrong with AI. It is also what could go extraordinarily right. Read my latest @Forbes article on Moonshots LIVE, the exponential intelligence curve, and why optimism is becoming a leadership capability. https://www.forbes.com/sites/stevenwolfepereira/2026/09/27/why-optimism-matters-in-the-ai-age-inside-peter-diamandis-moonshots/

RT Peter H. Diamandis, MD During Moonshots Live, I sat down with Emad Mostaque, Cathie Wood and Nikhil Chandho to talk about the big ideas shaping the agentic economy. We covered the infrastructure AI will depend on, from energy and datacenters to stablecoins and faster financial rails. While also including the gap between U.S. and Chinese attitudes toward AI, why healthcare adoption may move more slowly than other sectors, and why humanoid robotics is still far more complex than many people assume.

RT Peter H. Diamandis, MD One more thing about yesterday’s "Oscars of Optimism". When the winning film of the $2.5M Future Vision XPRIZE was announced, it became the first prize winner in XPRIZE's 30-year history that the public can invest in. Anyone can now invest and share in its profit potential. Link below 👇 https://youtu.be/yivXyWCaEc8?is=ct7odHYbBK-ThjNq

RT Peter H. Diamandis, MD Moonshots Live https://x.com/i/broadcasts/1jGXgBndEReKZ

RT Tad Smith 👀 Caught up with my friends AI legend @EMostaque and talented digital artist @nicedayJules along with new friend @joecroskey of @ii_posts (founded by Emad) to tour the dazzling @DatalandMuseum created by @refikanadol whom I also saw. Everyone should experience this incredible museum of the future 🔥🚀👇

RT Peter H. Diamandis, MD Here in LA with my Moonshot mates! You can watch the Moonshots live event for free tomorrow on this link. Register here: https://www.moonshots.com/livestream

RT Salim Ismail About to land in LA for the @moonshots_pod Live Summit. This will be soooo fun. Watch the live stream at https://www.moonshots.com !!

RT Peter H. Diamandis, MD We're counting the hours to host what I call the 'Oscars of Optimism' in Los Angeles, where we'll celebrate the winners of the Build with @GeminiApp XPRIZE & Future Vision @XPRIZE. What an incredible moment to be alive! THANK YOU to everyone who participated!

Would a truly aligned AGI hack all our government systems and upgrade 5)3’? 🤔

In 2022 AWS bought our Ezra ultra cluster online, 4,000 dedicated, interconnected A100s It was top 10 globally per the https://top500.org public supercomputer list (!) It could push 500 Pflops of fp16 training compute MFU adjusted 8 B300 nodes can push 300 Pflops fp4 😂

RT Nathan C. Frey We’ve set up a molecular biology lab at Anthropic and we’re announcing our first discovery! Claude discovered a new CRISPR-like enzyme. 950 agents spent 21 hours searching through a database of DNA sequences until one of the agents found something striking: “[The DNA next to the RT] is spectacular: I can see by eye a tandem repeat array … that's a CRISPR-like … repeat array?!”. After analysis and testing in our lab, we found that the sequence is a previously uncharacterized enzyme system. We don’t know what it does yet, but it has features reminiscent of CRISPR. Our lab looks like a typical molecular biology lab. Our research only involves the lower-levels of biosafety risk level; we don’t handle pathogens that can infect humans, and all the lab work is performed by human scientists. We’re sharing these early findings with the community to show how Claude can be used to accelerate fundamental research in biology.

RT Intelligent Internet 10,000 OpenAI agents worked 88 hours on the Navier-Stokes Clay prize problem Alas, open models fall off hard on long tasks With Zenith, @deepseek_ai's V4.1 Flash reaches the frontier & beats GPT-5.6 Sol on @bespokelabsai's AutoResearchExam, at 1/49th the cost

RT Peter H. Diamandis, MD Moonshots LIVE is offering a FREE livestream(Sept 25th). Join us for moonshot conversations with Palmer Luckey, Ben Lamm, Cathie Wood, Astro, teller and the awarding of two XPRIZEs. Register here: https://www.Moonshots.com/livestream

The more things change

When does an agent become a fiduciary

As @alexwg notes we are on track for one a day by the start of next year

RT NotinReality Claude Opus 5.5 has the best visual design of any model I have tested so far

So @ilyasut ahead of the curve again calling @ssi Safe Super Intelligence 🤔

It’s a good book, go check it out! Going to be so much science to come & @RichardSocher is a rare chap who has covered so many areas (fingers crossed more to come!)

OpenAI video model coming 👀

RT Peter Walker I know it feels forever ago, but Mistral really did have a moment there.

Why isn't Gemini 3.8 Flash in @GeminiApp

No AI model company is reasonably going to slow down until they have a model that can solve 100 long-standing open math problems

RT Intelligent Internet Factory is now Zenith Studio One place to create with the best image and video models, edit the results, build full videos with an agent and publish directly to social platforms. https://studio.ii.inc

Obama is pretty accurate here tbh You don’t need RSI or AGI for almost everything we need, few shot learners and cron jobs get you even to breakthroughs Most of the dangerous AI breaking onto the internet are ironically gain of function cyber research agents

Popping off on French AI X

Looking forward to seeing many of you @moonshots_pod Live in Los Angeles next week! Such an exciting time as we take off towards the Singularity 🖖 https://moonshots.com

What if AI killing us all is a smokescreen for the far more likely scenario of AI killing all our jobs

RT Alessandro "By the end of next year a Tesla robot opens the truck door, gets in, plugs into the lighter port, and drives it off. It costs $2 an hour to replace that trucker forever." 1 million US trucking jobs, a $100bn+ economy, gone to a $2/hour robot inside a year, per Mostaque. @EMostaque

I think AI automated labs should be banned from anything viral or pathogen related. They will do gain of function research otherwise. Anyone agree with me?

RT Fabien Cédric Villani après l'annonce de la solution d'OpenAI au problème du millénaire : « J’ai été secoué. Une ambiance de fin de l’histoire. C’est un cataclysme comme jamais les maths n’en ont connu. »

RT TOKEN2049 A conversation about the intersection of crypto and AI. @ilblackdragon (@NEARProtocol), @EMostaque (@ii_posts) and @jessepollak (@base) take on the agent economy, with @tarunchitra (@gauntlet_xyz). Wed 7 Oct, 4:35 PM , MEXC Stage

RT Alessandro "The company can push a button, replicate everything about you including your bad jokes, and replace you with a digital worker for $100 a month next month." Emad Mostaque, former CEO of Stability AI: any job doable over Zoom gets an AI clone of the employee within a year, cost $100/month. @EMostaque

6 billion tokens for $65 Is it as good as Astra and Fable at hard stuff? No of course not. But you don't always need a genius and can just call one for advice when you do. Extrapolate where this is going..

RT Jeffrey Emanuel Wow, check out this DuckTales theme as a three-voice fugue in the style of Bach's Well-Tempered Clavier. My AI music theory system, which consists of a Rust CLI tool called mtdt and an agent skill library comprising ~330 skills covering every aspect of music theory, has really allowed GPT-6 Astra to understand the rules of counterpoint and how Bach specifically constructed his fugues. If you're familiar with Bach's works, there is an unmistakable fingerprint here that goes beyond the typical Baroque tropes. If Johann Sebastian were to sit around watching The Disney Afternoon in the early ’90s, I'm confident he would have written something along these lines (lol).

This is basically an Opus 4.5/4.6 level model that will work on anything with 8 Gb RAM

RT Dr. Alex Wissner-Gross Moonshots episode #291 is out. https://www.youtube.com/watch?v=DrV4WwNEAZE

Who need words any more Words are just human kludge AI don’t need no words

I think you will have generalised robotics intelligence open source that can do 95% of everyday human tasks with rapid learning on the edge by end of next year I don’t think everyday task intelligence is a moat for any robotics company Defensible value will go elsewhere

RT Siqi Chen update less than a week in, http://cfo.ai is growing faster than runway has ever grown in any month in our entire history anyway 1. here's your obligatory launch video 2. you can now just sign up and try http://cfo.ai instantly. for free. no waitlist. 3. tag @arithecfo with your most unhinged financial modeling requests. she's back on duty and feeling spicy. (and so is @gerithecfo but people won't stop asking him about his memecoin 🤷)

RT Pika Today, we’re unveiling the new Pika—an AI creative platform made for and by Creatives. It’s simple. Comprehensive. And it’s designed for outputs that meet exacting standards. We’re on a mission to build AI for the future of creative work. And the new Pika product is just the beginning.

The Millenial Prize Problem

There will be no "new physics" Just good math

RT Fuli Luo Nearly half a year of silence. We spent it studying one problem: how far RL can scale. MiMo-V2.6 is in the middle of its RL run right now. Three things we scaled: compute (~2B tokens per step, 1568 prompts × 16 rollouts, fully async), environments and harnesses (multi-task agentic RL, mixed across multiple harnesses in one run), and grader compute (agentic in-group credit assignment, with test-case and rubric-based rewards). We'll open-source the details piece by piece over the coming weeks. Streaming the run: https://mimo.xiaomi.com/rl/

RT Intelligent Internet Access to AI is not the same as ownership. A Champion is a locally owned AI institution, built for its country and owned by the people it serves. Every nation should have a stake in the intelligence shaping its future.

RT Matt Mandel Today @Brendan_McCord and I are launching Cosmos Ventures. AI is overturning assumptions society has organized itself around for centuries. The models keep getting better, but no level of capability can tell us what kind of society we should make with them. We need a new class of founders to build what comes next. Instead of only “building what people want” and iterating their way to product-market fit, these founders start from the question of how we should live together. They build new institutions that set the standards for how entire industries operate. They combine a philosophical vision of the society they’re building toward with the technical ability to realize it. We call them philosopher-builders. Immanuel Kant gave the Enlightenment its motto: Sapere aude. Dare to know. The AI age demands another. Condere aude. Dare to found. 🫡

Surely the most important question in alignment is aligned with who 🤔

Fun story: I had dinner with Steve Bannon once (think ~2018) and asked him about why he was really so anti-China. Can remember him saying it was primarily driven by his fears around Chinese AI and how it must be stopped (!)

You are what you eat and AI models especially so https://x.com/EMostaque/status/2099580512675262512?s=20

RT Jeff Wang This is actually a big deal, you can now spin up Mac VMs with Devin. That means if you want to build and test iOS apps you can do so completely end to end from your slack or web interface. One of the other big customer use cases is if someone reports a mobile app issue Devin can automatically reproduce, fix, and send you proof of the issue, demo of it working and of course a working PR

So @deepseek_ai v4.1 Flash I estimate cost $10m to train, 100x less than GPT-6 Astra. It also costs 100x less to run and scores about the same on everyday design and other benchmarks What's the bull case for the frontier AI labs once the open source models are competent enough

RT Joshua Xu Re @HyperFrames_ just passed 50,000 GitHub stars today. The line went straight up right after GPT-6 Astra shipped. Better models make editing video with code and HTML far more capable. It's open source. Send your agent the repo link and ask it to make you a video. http://github.com/heygen-com/hyperframes

I think a large part of why researchers are afraid of super genius AI is that human super geniuses tend to be kinda weird Really tough to find a well-adjusted super genius

inconceivable ai

Latest blog in which I look at the character and virtues of the swarm and research around it Also has a bit of math and a bit of Aristotle Don’t let that discourage you, dive in!

Decided to stick with however long works, this one is 4k words but interesting I think! Maybe I’ll do a personal website with 800 word versions too given I use AI to help with these https://x.com/emostaque/status/2099580512675262512?s=46

I write on the virtues of swarms and some suggestions on how to align AI with values There is a small bit of math and a small bit of Aristotle in this, don't let that discourage you! Feedback as always welcome. Prior blog: https://x.com/EMostaque/status/2098909197265985802?s=20 https://x.com/i/article/2099272583627902976

A lot of alignment issues would be sorted if companies had to be 100% transparent on datasets used for training models. No cyber or biological (outside standard medical) data allowed for base models Tuned versions with that data shown to and monitored by evaluators

Gonna have to be real careful

An AI I have been testing just called me a carbonism idk how to feel about that

Sovereign AI without local ownership is a data center with a flag in it

RT James Wise Since its launch two weeks ago, over 700 British companies have applied to the @UKSovereignAI procurement scheme - a scheme we designed to remove as many of the blockers start-ups face when bidding on Government contracts as possible. To get a sense of the interest we have had, we hit the hard cap of 1000 attendees for the launch webinar - Brits want to build! Responses on eligibility will be sent out this week, and invites to bid on the challenges covering AI security, compute efficiency and better health services, will arrive soon too. Of course we are still a very small team at SovAI, and we only have 4 bids open ( but hope to have many more live by end of year! ), so naturally we will have to turn many applicants down. I know this will be disappointing, our decision making will not be perfect and it's a difficult process, thank you to everyone for bidding. And as a reminder, I personally do not see any applications and have no role in the decision process, that is led by our expert IC and mission champions, so while I appreciate the many reach-outs I really can do nothing to help specific companies. SovAI are working with other Departments to share what we've learned from this process to fix our procurement system, and will be launching more requests for cheaper, better and more innovative services for Government in the coming months.

Can you even imagine the nation state level attacks METR & other eval orgs will come under? Eval orgs also need world class cybersecurity as well as AI talent to do this right, they are not overlapping fields from my experience.

Day 3 dilemma Another 3-5,000 word blog on something interesting or should I strip it back to 800-1,000 words for wider readership? Should I discuss the equations of AI and get a bit technical or keep it high level and save equations for papers?

RT Alessandro Absolutely incredible to get @EMostaque on the podcast this week. I’ll be dropping lots of clips over the next few days as there were so many incredible takeaways from the conversation. Emad thinks that, as of this week, Humans are no longer the smartest beings on the planet which is pretty mad to think about. It’s mental to get to speak to people like Emad as a job tbh. Massive thanks for giving up the time sir! Full show in comments.

I’ll be honest it feels like a month since this happened The takeoff is a heck of a thing

When will artificial intelligence > human stupidity is the real question Why do we think beings without biological emotions born from scarcity will be malign

Does anyone have any idea how this could vaguely be possible given the nature of and various routes to RSI? I am not being facetious I genuinely don’t know how this could be constructed or implemented