Ilya saw reasoning, or more like reasoning potential. I think it’s obvious by now. It’s not obvious what they did after that. They needed to create a strategy to fight Google and its enormous compute. Since that time, they have focused on efficiency and created tons of diverse…
I am contemplating this statement a lot recently. It feels possible, and for this reason I am not sure what to think about it. Training compute will increase only a few times before the end of 2026, so there is that.
On the other hand, there is anecdotal evidence that GPT-5 Pro…
I am contemplating this statement a lot recently. It feels possible, and for this reason I am not sure what to think about it. Training compute will increase only a few times before the end of 2026, so there is that.
On the other hand, there is anecdotal evidence that GPT-5 Pro…
Maybe I’m mistaken, but this is mostly “just” SSM like gains? Obviously important for agents, though, but not at this size, so we don’t know anything practical, I guess.
Google is starting to dominate in some visual areas, likely because of its “infinite” data. Gemini 2.5 is a good model but lacks agentic behaviour, so they still need to build RL pipelines like everyone else. If Gemini solves agency with Gemini 3, TPUs could make them almost…
Google is starting to dominate in some visual areas, likely because of its “infinite” data. Gemini 2.5 is a good model but lacks agentic behaviour, so they still need to build RL pipelines like everyone else. If Gemini solves agency with Gemini 3, TPUs could make them almost…
Ok, so this is so huge that I'm not sure what to think about it. It would be impressive if this model were 2 or 3x faster, but 53x faster generation is confusing, and my first thought is where is the catch or if this scales?
Ok, so this is so huge that I'm not sure what to think about it. It would be impressive if this model were 2 or 3x faster, but 53x faster generation is confusing, and my first thought is where is the catch or if this scales?
What is most surprising is that it supposedly requires a consumer GPU only. Even if it's a 5090 now, it's in the range for reducing it to something average. I'm not sure how world persistence and introducing custom mechanics will work here without very complex text/visual prompts…
What is most surprising is that it supposedly requires a consumer GPU only. Even if it's a 5090 now, it's in the range for reducing it to something average. I'm not sure how world persistence and introducing custom mechanics will work here without very complex text/visual prompts…
Introducing DeepSeek-V3.1: our first step toward the agent era! 🚀
🧠 Hybrid inference: Think & Non-Think — one model, two modes
⚡️ Faster thinking: DeepSeek-V3.1-Think reaches answers in less time vs. DeepSeek-R1-0528
🛠️ Stronger agent skills: Post-training boosts tool use and…
20K Followers 9K FollowingProgramme Director @ARIA_research | accelerate mathematical modelling with AI and categorical systems theory » build safe transformative AI » cancel heat death
6K Followers 7K FollowingFOLLOWS YOU 🫵 https://t.co/F7MzDOTC1k
ML/AI, R&D eng, quant trading, ASR in noise, TTS.
OPEN weights, thoughts, ... AGI, ASI - open AI computation for */acc—NOW 🥰
3K Followers 2K FollowingHumanist technologist and AI optimist. Currently CTO at @welcomeaccount_. Building for an inclusive economy through #AI, #MachineLearning, and #Tech4Good
88K Followers 8K FollowingCompiling in real-time, the race towards AGI.
🗞️ Don't miss my daily top 1% AI analysis newsletter directly to your inbox 👉 https://t.co/6LBxO8215l
24K Followers 4K FollowingTurnaround CTO, advisor, and startup vagabond. Former head of AI @NASA CAS and tech wonk for (Obama) @WhiteHouse, DOD, and DOJ. Tweets are my own.
489K Followers 146 FollowingNobel Laureate. Co-Founder & CEO @GoogleDeepMind - working on AGI. Solving disease @IsomorphicLabs. Trying to understand the fundamental nature of reality.
22K Followers 1K FollowingI make games with AI. Product Designer, Co-founder at https://t.co/C6xWXxJrsL & https://t.co/5jKwzHKucU. Founding Designer at RedMart, acquired by Alibaba.
19K Followers 1K Followingferal attention researcher ᐧ strangely earnest ᐧ I teach Alexander Technique, which helps you get out of your own way → https://t.co/e0qcOYJBfG
10K Followers 131 FollowingBuilding super-tiny AI models that (hopefully) think • YC S25 • Made @DiffusionBee • Previously: AI research @Microsoft, @CarnegieMellon, @Meta
55K Followers 2K FollowingHead of Design @Cursor_ai. Early @NotionHQ, @Stripe, built startups. I make a world where anyone can make software. Aspiring k-pop idol.
6K Followers 7K FollowingFOLLOWS YOU 🫵 https://t.co/F7MzDOTC1k
ML/AI, R&D eng, quant trading, ASR in noise, TTS.
OPEN weights, thoughts, ... AGI, ASI - open AI computation for */acc—NOW 🥰
22K Followers 747 FollowingBootstrapped nomad for 8+ years 🎒 in 50+ countries 🗺️ w/ @jamesivings
Unsubscribe from emails 💌 https://t.co/8TZvCY7i1u
Boilerplate for AI SaaS 🤖 https://t.co/tU9DVVSAKM
26K Followers 1K FollowingDigital Nomad for 8+ years, bootstrapping products around the world! 🌏
🙅♂️ Easily unsubscribe from emails - https://t.co/5bc2Kq5mIR
25K Followers 927 FollowingShipping like a machine 🤖 while I travel the world 🌍️
🔥 https://t.co/myyLdxVzcg - Grow faster on X ($7K/m)
🚀 https://t.co/5fSvUeZJ8z ($38K)
🤝 https://t.co/l4mcaViGk7
🕵 https://t.co/Cwvsl9cPsS
4K Followers 894 Followingsolve cooperation, use it to solve everything else.
collective intelligence research @ midjourney
longer essays at https://t.co/5w7LaGnW6l
3K Followers 2K FollowingHumanist technologist and AI optimist. Currently CTO at @welcomeaccount_. Building for an inclusive economy through #AI, #MachineLearning, and #Tech4Good
88K Followers 8K FollowingCompiling in real-time, the race towards AGI.
🗞️ Don't miss my daily top 1% AI analysis newsletter directly to your inbox 👉 https://t.co/6LBxO8215l
539K Followers 18K FollowingThe best from AI community | Ex-Microsoft, Rackspace, Fast Company | Wrote eight books about the future | Silicon Valley robots, holodecks, BCIs, and startups.