It’s been 15 years since I last wrote an academic paper and I didn't expect the next to be this fundamental. Today we published a research paper with @openservai that represents a major breakthrough in how we think about AI reasoning. For years, we’ve been told that better AI reasoning means bigger models, longer chains of thought, and higher costs. We accepted that trade-off almost by default. This work questions that assumption. With BRAID's structured, bounded reasoning, we’re seeing up to 74× efficiency gains and ~30× better performance per dollar without losing accuracy.

Turns out @UberEats Germany customer support rivals RyanAir's, basically next to scam. I will be cancelling my Uber One subscription and will never order again. Shame on you.

European Commission is so wrong about “competition”, “innovation”, and AI, and Apple, I am appalled. Like, who are these people? Did they _ever_ innovate on anything? Did they compete on anything? The DMA mental model is so wrong, that they single-handedly leave EU out of the world’s largest innovations. Doing *exactly* the opposite of what’s intended. Just like GDPR, just like the rest of European politics in the last decade. It makes me question the wisdom—these people are clearly misinformed and lack basic tenets of judgment and wisdom. These decisions aren’t even intelligent. You are running a continent. You have to foresee the outcomes of your actions exactly. I’m very sorry for Europe…

AI companies were touting the cost of intelligence getting smaller following Moore's Law, and even claiming that the prices are halving every 6 months... now what we see instead is prices are doubling every 6 months. This is unsustainable.

my thoughts on $SERV using blockchain: longterm goal is to put all agentic reasoning on chain, this greatly enhances the audibility angle tokenization also allows the common man to gain access to the project much earlier than would be possible if they took the trad VC route, and that’s honestly something this team cares about I’ve talked with some of them directly and tbh they are the kind of people who only want to build if their work benefits humanity, this isn’t just a quick money grab for them, it will be their life’s work they easily could have gone the VC non-crypto route, but they honestly believe crypto rails are superior and didn’t want the profit constraints VCs would force at its core, this is an AI research lab and @dashersw is a true visionary, he is inventing a better AI and I sense that he prefers the freedom to work outside of the incessant VC demand for profits over people ultimately, this mindset is why @openservai will win and the big labs wont

One of the, if not THE smartest minds in machine learning is giving an #UncTalk in the $SERV Telegram again. 🤯 This team is different! @openservai t.me/openservai

Just use SERV Reasoning.

What is happening at @openservai on a beautiful Thursday afternoon Our CTO @dashersw is giving the community a deep technical lecture on how LLMs evolved, why the current architecture is broken and why $SERV is building a different path And apparently this has already been going for around an hour This is the kind of thing I love to see Not just hype Not just price talk He's explaining the history, the architecture, the weaknesses of current LLMs and why SERV Reasoning can become such an important unlock When a team takes the time to educate its community at this level, it says a lot @openservai is not only building They are making sure people understand why what they are building matters

Another day, another build.

Yazılım öğrenmek kolay mı sandın? Sosyal medya yorumlarına cevap veriyorum | 29 Mayıs Cuma 22.00 x.com/i/broadcasts/1RJjppDrv…

What is Raison? Well I’ll tell you. Raison is another piece of the complete stack that @openservai is building to make #AiAgents actually scalable for enterprise large and small. Raison is another cutting edge product that comes out of the genius combination of @dashersw unique mind and the $SERV team of cracked devs. Raison is the manifestation of Armagans statement, “we are looking 10 years out” “we have many products in development”. Raison is an example of earning trust with the Reasoning Framework then upselling the entire stack. Raison is proof you are asleep and need to wake up to @openservai being a Trillion dollar monster larping at $50m.

You don't need a more expensive model. You need better reasoning. Every frontier LLM got smarter with SERV: - smaller ones match frontier performance - frontier ones go past what they were built for The question isn't which model to choose - it's what engine it runs on.

None of this happens without our sponsors: @coyotiv , @fal , @xata , @EDBPostgres , Mobileitm . A special note to @coyotiv and @dashersw you weren't just sponsors, you were part of building @divaconference with us. 💫 Thank you. 🎉

SERV is now powering robots. That's it. That's the tweet.

Another production agent left Gemini for SERV. SERV is inevitable.

Robotics is another industry running on SERV. @Roba_Labs compared SERV to Claude across 40 tasks on a Unitree G1 Humanoid: file edits, sim-to-real workflows, robotics asset packaging. "SERV matched Claude's output quality - and cut our AI costs by over 80%. That benchmark result changed our roadmap. serv-standard is now the default model in ROBA Studio." - Farid Hossain, founder of Roba Labs

Huge appreciation to the @openservai team for their full support throughout the process. After 4 years of working with AI models, I can honestly say SERV Reasoning is one of the best combinations of performance and cost efficiency I’ve used. Better reasoning. Faster responses. Lower cost. A significant step forward for XONA.

we're witnessing a clear trend: a multi-trillion dollar market running on infrastructure that's getting more expensive every cycle just read Gartner's latest AI forecast: - $2.59T market in 2026 ( 47% YoY) - $3.5T in 2027 while frontier model prices keep climbing: GPT-5: $10/M -> GPT-5.4: $15/M -> GPT-5.5: $30/M it's clear enterprise will be looking for cost-savings while adopting AI at large scale. this puts ethereum:0x40e3d1a4b2c47d9aa61261f5606136ef73e28042 at the center of the AI agent economy - delivering up to 100x better performance-per-dollar in enterprise benchmarks, with no vendor lock-in. with a single-line swap, you plug any frontier model into SERV Reasoning and the economics flip in your favour: agents become not just efficient, but also reliable and fully auditable. it's why most enterprises in SERV Reasoning Private Beta are already moving over, once they try it UAE gov is using SERV tech in production; Akretic (security layer for finance, healthcare, government, defense), treasury infra (BILLZ), agentic OS for the food industry (GastroSight), agent compliance layer (ThoughtProof) - AND MANY MORE - all switching their entire stacks private beta already crossed 100K requests last week. incredible foresight PMF from @openservai

I now have a pretty good evidence that gpt-image-2 is an HTML renderer. On the left: actual HTML implementation of a slide. On the right: gpt-image-2 output. It consistently nails it down every single time.

Asking two agents in different harnesses to debug your code (from andirockk on IG)