Content History
We bumped up the testing component of II-agent so it has better outputs than peers In normal software development you spend way more time testing than coding Going to experiment with this
We put 4 leading agents to the test with a 3D Runner game 🎮 Same prompt. Same concept. Very different execution. Which one would you play?
View quoted postDeepSeek’s fund beat all of these…
The Top Hedge Funds of 2025 Apis Capital led the pack at 55.1%, with several Tiger Cubs, including Tiger Global, Lone Pine, and Maverick, making the cut.
RT Intelligent Internet Give it an image and a direction. 20+ styles, precision edits, model switching, exact ratios — all in one flow. II-Agent Advanced Image Suite Visuals, perfected. Original tweet: https://x.com/ii_posts/status/2009680394405212352
RT Dan Shipper 📧 NEW: i wrote a complete technical guide to building agent-native software (co-authored with claude) it covers: - the five pillars of agent native design (parity, granularity, composability, emergent capability, self-improvement) - files as the universal interface - agent execution patterns with code samples - mobile agent patterns - advanced patterns like dynamic capability discovery if you want to take full advantage of this moment, it's worth your time: https://every.to/guides/agent-native?source=post_button Original tweet: https://x.com/danshipper/status/2009651408144835021
RT Intelligent Internet We upgraded intelligence without upgrading the system around it. That mismatch is the story of our time. Full audiobook out now on Spotify and Apple Podcasts👇 Original tweet: https://x.com/ii_posts/status/2008963492598968564
Incoming Amusingly the core of the paper can be summarised in two lines But a very important two lines it is
RT Armand Domalewski wow https://www.nytimes.com/2026/01/02/world/asia/china-ai-cancer-pancreatic.html?unlocked_article_code=1.B1A.iv9-.WzkJolxDYex4&smid=nytcore-ios-share Original tweet: https://x.com/ArmandDoma/status/2007697062926921767
Great job by Jian Yang & co!
2026: A new AI lab out of China claims 81.4% on SWE-Bench-V and 54.2% on BigCodeBench with a 40B coding LLM. IQuest Labs, like DeepSeek, comes from a quant fund Ubiquant in China with $8B AUM. Jian Yang, the leading contributor, seems to have been on the Qwen2/3 team as well!
RT Intelligent Internet Merry Christmas from Intelligent Internet 🎄🎅 Original tweet: https://x.com/ii_posts/status/2004335976362164697
My tribute to Ramunajan & a challenge to math lovers… This is accurate to ~6ppm (!) Can you get closer elegantly ^_^ Have fun!
I find automated math proving with AI not hugely interesting What is most interesting is how we can leverage AI to enable more intuition and inspiration like Ramanujan Save the proofs for later, dare to explore and do what we do best while the GPUs do what do they best
View quoted postRT João Batalha von Neumann gave us "zero-sum" in 1944 and it's one of my favorite examples of linguistic compression. Two words that mean “strictly competitive, fixed pie, stop looking for win-wins.” The term's real gift: it reveals the default human assumption that life is chess and value can’t be created—only transferred. Once you have the label, you start noticing how often that’s wrong Original tweet: https://x.com/joao_batalha/status/2003540220352897031
I find automated math proving with AI not hugely interesting What is most interesting is how we can leverage AI to enable more intuition and inspiration like Ramanujan Save the proofs for later, dare to explore and do what we do best while the GPUs do what do they best
Celebrating the combination of intuition and experimentation in the mathematical works of Srinivasa Ramanujan on the anniversary of his birth #WithWolfram https://writings.stephenwolfram.com/2016/04/who-was-ramanujan
RT Wolfram Celebrating the combination of intuition and experimentation in the mathematical works of Srinivasa Ramanujan on the anniversary of his birth #WithWolfram https://writings.stephenwolfram.com/2016/04/who-was-ramanujan Original tweet: https://x.com/WolframResearch/status/2003152362781757711
RT Fermat's Library Srinivasa Ramanujan was born 138 years ago today. He said his formulas came to him as revelations from God. Original tweet: https://x.com/fermatslibrary/status/2003025304147616163
RT Ethan Mollick Surprisingly rapid & high Ai adoption by doctors: 67% use it daily, 84% says it makes them better doctors, 42% says it makes them want to stay in medicine more (10% said less). A lot of the use cases appear to be administrative and research assistance. https://2025-physicians-ai-report.offcall.com/ Original tweet: https://x.com/emollick/status/2001061282485547116
The "thinking" bar for GPT 5.2 Pro has got a lot better, you can see it looking at the pages etc of what goes in, interrupt it etc However it remains on the web version only (along with the ability to flip to high reasoning), makes the app version not v useful :(
The Last Economy is number 1 on Amazon's Best Sellers for AI with a pretty good ratings ✍️ No publisher or big campaign, also available for free on the website and paperback just dropped 📖 Thank you all for the support & the next part is nearly done..🚀
Economics was built for scarcity. The Intelligence Age is about abundance. Today we release our book, The Last Economy, introducing Intelligent Economics, a unified theory for this new era. 🧵
View quoted post68% on SWE-Bench Verified on just 24b! Laptop class 72.2% on 124b 👏 Great job @MistralAI team, looking forward to trying it out
Introducing the Devstral 2 coding model family. Two sizes, both open source. Also, meet Mistral Vibe, a native CLI, enabling end-to-end automation. 🧵
View quoted postRT Intelligent Internet The Last Economy by @EMostaque is an Amazon Best Seller 📚 #3 Business Economics (kindle) #3 Finance (kindle) #7 Computing Your guide to the AI economy & what comes next Thank you for reading, tell your friends. Now also available in paperback 🎁 Original tweet: https://x.com/ii_posts/status/1998411140502556694
RT Umut Gunbak fal Startup Program in Europe! Looking for ambitious builders We’re launching our first startup program in Europe to help generative media teams build and scale with @fal Apply for one of three packs: • Starter Pack: $250 fal credits • Startup Pack: $1,000 fal credits • Scale Pack: $5,000 fal credits If you’re just getting started, grab the Starter Pack. As you grow, you can apply for the next tier. Let’s build the next decade of generative media together. Link in the replies Original tweet: https://x.com/umutgunbak/status/1997933230243058090
RT Intelligent Internet Terminal Bench v2 just got officially verified by @alexgshaw II-Agent is now Top 2 overall and #1 single-model agent Open Source AI, top-tier performance 🚀Emad: Buried the lede a bit but our fully open source II Agent framework is now state of the art in Terminal Bench 2 using just Gemini 3! Congrats to team for amazing work & more coming the pipeline The best agents will be open source so we can all have the sovereign AI we deserve Link: https://x.com/EMostaque/status/1991186525975970072
Humanity as the biological bootloader of AGIJames Moore: Bootloader?!! Link: https://x.com/jamesdmoore614/status/1992679980816327049
This is more than all but one LLM and probably more than used to train every generative video model ever Still a way to go!Cosmic Marvel: Disney has confirmed that the total budget for ‘ANDOR’ is over $705M. The highest-ever for a Star Wars project. (via https://www.forbes.com/sites/carolinereid/2025/11/22/disney-spent-more-on-andor-than-any-of-its-star-wars-movies/) Link: https://x.com/cosmic_marvel/status/1992383951135203578
A billion humanoid robots and a bad firmware update is one of the most reasonable human (Doom) scenariosMike Kalil: Figure AI, an Nvidia-backed developer of humanoid robots, was sued by the startup's former head of product safety who alleged that he was wrongfully terminated after warning top executives that the company's robots "were powerful enough to fracture a human skull." Robert Link: https://x.com/mikekalilmfg/status/1992269576155922863
The most interesting thing testing Gemini 3 Pro has been how *efficient* it is from tokens to tool calls The intelligence per token of models is increasingly rapidly even as prices fall, its quite somethingLogan Kilpatrick: Introducing Gemini 3 Pro, the world's most intelligent model that can help you being anything to life. It is state of the art across most benchmarks, but really comes to life across our products (AI Studio, the Gemini API, Gemini App, etc) 🤯 Link: https://x.com/OfficialLoganK/status/1990813077172822143
In a year or so this level of performance should be available on a 32b dense model (K2 is 32b active) at a cost of < $0.2/million tokens I don't think folk have that in their estimatesKimi.ai: 🚀 Hello, Kimi K2 Thinking! The Open-Source Thinking Agent Model is here. 🔹 SOTA on HLE (44.9%) and BrowseComp (60.2%) 🔹 Executes up to 200 – 300 sequential tool calls without human interference 🔹 Excels in reasoning, agentic search, and coding 🔹 256K context window Built Link: https://x.com/Kimi_Moonshot/status/1986449512538513505
A note on costs/compute Base Kimi K2 model used 2.8m H800 hours with 14.8 trillion tokens, about $5.6m worth Details of post training for reasoning not given, but it is likely max 20% more (excluding data prep!) Would be < $3m for sota if they had Blackwell chip accessKimi.ai: 🚀 Hello, Kimi K2 Thinking! The Open-Source Thinking Agent Model is here. 🔹 SOTA on HLE (44.9%) and BrowseComp (60.2%) 🔹 Executes up to 200 – 300 sequential tool calls without human interference 🔹 Excels in reasoning, agentic search, and coding 🔹 256K context window Built Link: https://x.com/Kimi_Moonshot/status/1986449512538513505
Congratulations to @Kimi_Moonshot for achieving state of the art on many benchmarks & open sourcing the model! The gap between closed & open continues to narrow even as the cost of increasingly economically valuable tokens collapses K2 has its own unique vibe too, try it out!Kimi.ai: 🚀 Hello, Kimi K2 Thinking! The Open-Source Thinking Agent Model is here. 🔹 SOTA on HLE (44.9%) and BrowseComp (60.2%) 🔹 Executes up to 200 – 300 sequential tool calls without human interference 🔹 Excels in reasoning, agentic search, and coding 🔹 256K context window Built Link: https://x.com/Kimi_Moonshot/status/1986449512538513505
RT Dheemanth Reddy here is maya1, our open source voice model: We’re building the future of voice intelligence @mayaresearch_ai team is incredible; amazing work by the team. remarkable moment.
