NAll news

All news

Wednesday626 items

  1. Ayi AI Notes
    AI score75

    Axiom says AxiomProver agent swarm completed Lean formalization of the Classification of Finite Simple Groups theorem in 24 days

    Axiom announced the completion of a complete Lean auto-formalization of the Classification of Finite Simple Groups theorem; the proof spans about 15,000 pages and was completed by over 100 mathematicians across generations. According to the account, the AxiomProver agent swarm produced about 50 million lines of Lean code in 24 days and fixed logical gaps in the original work.

    Quoted postAxiom · @axiommathai

    The largest math proof in human history is now computer-verified. Today, Axiom announces a complete Lean auto-formalization of the Classification of Finite Simple Groups, the "periodic table of symmetry." About 15,000 pages, built by over 100 mathematicians across generations. https://t.co/Y1P26abyPQ

    Same story · 6 reports

    #Papers/Research#Agents#Reasoning

  2. Larsen Cundric
    AI score26

    Grok bot version 0.77 supports multiple questions on a single card

    Grok's bot 0.77.0 is now stable and rolling out, adding permission settings for auto-review rules, the ability to prioritize payment methods in the vault, and the ability for bots to ask multiple questions on the same card, reducing back-and-forth communication.

    Quoted posteric zakariasson · @ericzakariasson

    grok @bot 0.77.0 is now stable and rolling out! - permissions settings for your auto-review rules - pick which payment method bots use first in vault - bots can ask several questions on one card full changelog: https://t.co/dyBGOXTxKP https://t.co/ottBxbgN4u

    Same story · 2 reports

    #Product updates

  3. indigoProduct release
    AI score65

    Google Genie 4 launches, built on Gemini Omni

    Google Genie 4 launches, built on Gemini Omni, and can present realistic details in spaces and specific themes. This version features advanced physics effects, fully real-time control, support for ambient sound effects, music, background noise, and interactive dialogue, and uses Gemini reasoning to plan multi-step events to maintain coherence as scenes evolve.

    Same story · 28 reports

    #Product updates#Multimodal#Google

  4. KazikeModel releases
    AI score79

    Google releases the world model Genie 4, adding native audio and real-time control

    Google has released the next-generation world model Genie 4, which improves on Genie 3 in image quality, motion, and physical interaction, and adds native audio generation that can synchronously generate ambient sounds, musical sounds, and character interaction dialogue. Genie 4 supports keyboard control of characters or the camera, and can also modify weather, add objects, and change events using natural language during runtime. Google has also integrated Gemini into the generation process as a guiding agent, responsible for observing the frames, breaking down goals, and checking frame by frame, while Genie is responsible for generating visuals, sound, and dynamic changes.

    #Model releases#Multimodal#Google

  5. MiniMaxFirst-party
    AI score20

    MiniMax and Fireworks AI to Discuss M3.1 Flash in Livestream Tomorrow

    MiniMax officially announced that it will co-host a livestream with Fireworks AI on October 28 at 10 AM Pacific Time, focused on MiniMax M3.1 Flash. MiniMax's Victor SuOrtiz and Fireworks AI's Sophia Yang will answer questions online about agent ideas and integration. The cited Fireworks post states that the livestream will introduce the model's new features, its intended design uses, and how to use it on Fireworks.

    Quoted postFireworks · @FireworksAI_HQ

    How good is MiniMax M3.1 Flash? @VictorSuOrtiz (@MiniMax_AI) and Fireworks' @sophiamyang go live Oct 28 at 10 AM PT to break down what's new, what it's built for, and how to use it on Fireworks today. Set a reminder: https://t.co/yIGF3knIkX

    #Industry news#Agents#English text is being prepared.

  6. Rohan Paul
    AI score34

    Pichai on using AI agents to keep up with Google's internal developments

    Rohan Paul relays what Sundar Pichai said on Alex Heath's podcast: Pichai said he has agents running internally that summarize information from chat groups on pre-training, mid-training, and post-training data; other agents are connected to telemetry data on compute usage, product areas, efficiency, and utilization, providing him with inputs and insights.

    Quoted postAlex Heath · @alexeheath

    How Sundar Pichai uses AI to run Google: "I have agents running internally for me... There are all these chat groups: pre-training, mid-training, post-training data, etc. It could easily take me an hour a day to catch up on what's happened in the last 24 hours, but now I have agents running that summarize and bring that information to me. I can still go deeper, and I do so often. Similarly, there are agents with telemetry into our compute usage, product areas, efficiency, and utilization, and they are giving me inputs and insights."

    #English text is being prepared.#Agents#Google

  7. Rohan Paul
    AI score75

    Bloomberg: OpenAI plans to purchase AI chips from Google

    Bloomberg reports that OpenAI plans to purchase AI chips from Google, incorporating its competitor's processors into its own computing infrastructure. Richard Ho, head of OpenAI's chip team, previously worked on Google TPU-related efforts. Oracle will also offer TPUs in its cloud services, and Google's chips are spreading as an alternative to Nvidia GPUs to external buyers and competitors' cloud platforms.

    Same story · 6 reports

    #Industry news#OpenAI#Google

  8. Rohan Paul
    AI score67

    Anthropic IPO Filing Sets Potential Market Size at $40 Trillion per Year

    According to a WSJ report, Anthropic estimated in its IPO filing that the economic value of all work by global knowledge workers in 2025 exceeds $40 trillion, and treats this entirety as its own potential market, higher than SpaceX's estimate of $28.5 trillion. Its argument is that frontier AI can boost the productivity of these workers, thereby expanding this portion of value.

    Same story · 121 reports

    #Industry news#Anthropic

  9. Artificial AnalysisAnalysis
    AI score65

    Artificial Analysis releases Intelligence Index v5, with 60% of evaluations based on private datasets

    Artificial Analysis releases Intelligence Index v5, adding two new benchmarks, FrontierCode 2.0 and Terminal-Bench-Science, upgrading HLE-Diamond, AA-Omniscience v1.1 and CritPt v2, removing GDPval-AA v2.1, SciCode and AA-LCR v1.1, and changing 60% of the index to evaluations based on private test sets or held-out answers. It also launches Progress Index v1, which uses a unified scale to track changes in model capabilities over time. The results show that Claude Opus 5.5 (max) ranks first with 58 points, followed by GPT-6 Astra (max, 57) and GPT-6.1 Sol (max, 55), while Gemini 4 Argon (high) scores 50; the gap between open-weight models and the leaders has widened from 3 points to 12 points, and among released weights, DeepSeek V4.1 Flash 1017 (max) and GLM-5.3 (max) lead with 32 points. The per-task cost of the top ten models differs by more than 13 times, with GPT-6.1 Terra (max) lowest at $0.96 and Claude Fable 5.1 (max) highest at $13.04.

    Same story · 8 reports

    #Evaluations / benchmarks#Agents#English text is being prepared.

  10. ModelBest OpenBMB
    AI score26

    MiniCPM-OCR Gets OpenVINO Day-0 Support

    ModelBest OpenBMB thanks the Intel Devs team for their support and announces that MiniCPM-OCR has landed on Intel hardware with Day-0 OpenVINO support. The cited Intel Devs post explains that MiniCPM-OCR can assess output reliability while extracting information from documents, and was optimized for Intel hardware at launch.

    Quoted postIntel Devs · @inteldevs

    Document AI is only as useful as the confidence you have in its results. MiniCPM-OCR helps extract information from documents while assessing output reliability. With Day 0 OpenVINO support, it's optimized for Intel hardware from launch. Try it today: https://t.co/Q2Q0uyNSPg https://t.co/OaTx2T7LDA

    #Product updates#English text is being prepared.#English text is being prepared.

  11. Kazike
    AI score31

    GPT-6.1 Terra hands-on test: faster and more economical at controlling browsers and computers

    After hands-on testing GPT-6.1 Terra, Digital Life Kazike switched his default model from GPT-6.1 Sol high to GPT-6.1 Terra xhigh for everyday tasks in which Claude controls browsers and computers. He rated the model as fast, with good performance and stability, and also very economical, making it suitable for handling routine control tasks.

    #English text is being prepared.#Agents#OpenAI

  13. Testing CatalogModel releases
    AI score71

    Google Releases World Model Genie 4, Supporting Dynamic World Modification and Native Audio

    Google has officially released the world model Genie 4, which it calls its most realistic and interactive world model to date. Users can dynamically change the world with prompts and generate scenes with native audio. The model will first be opened to Trusted Testers as a research preview, and will subsequently be made available via the Gemini API and AI Studio to Google AI Ultra subscribers and early access partners.

    Quoted postGoogle DeepMind · @GoogleDeepMind

    This is Genie 4, our most realistic and interactive world model yet. 🌐 Explore dynamic environments in real time to see how the scene responds, sounds, and how the people in it move, react, and speak.

    Same story · 28 reports

    #Model releases#Google#Multimodal

  14. Testing CatalogModel releases
    AI score60

    Microsoft MAI Image 2.7 Preview Launches on Arena and MAI Playground

    Microsoft has released the image editing model MAI Image 2.7 Preview, now available on Arena and MAI Playground. The model ranks 2nd on Artificial Analysis's AA-Image-Editing v2.1 leaderboard and 3rd on Image Editing Arena, a 45-point improvement over 2.6. Meanwhile, MAI-Image-2.6 and MAI-Image-2.6-Flash are now generally available in Microsoft Foundry.

    Quoted postMicrosoft AI · @MicrosoftAI

    MAI-Image keeps climbing! MAI Image 2.7 Preview is now #3 on @arena's image editing leaderboard. #2 by lab and ahead of GPT-Image-2. That's a +45 point jump from 2.6.

    Same story · 7 reports

    #Model releases#Image generation#Microsoft

  15. Testing CatalogProduct release
    AI score55

    Judgment Labs is now generally available, providing a continuous improvement toolchain for AI agents

    Judgment Labs has officially opened for general access (GA), positioned as a continuous improvement toolchain for AI agents; agent clusters can traverse production traces to discover failures, classify and investigate them, and test fixes. Its Judgment MCP can be used with Claude, Codex, Cursor, or any MCP client. Publisher Alex Shan said that long-running agent systems generate large amounts of trace data, which records user requirements, when agents go wrong, and how agents interact with the external world.

    Quoted postAlex Shan · @alexshander03

    Today, @JudgmentLabs opens for general access. We believe long running agent systems will generate more exhaust data than the world can currently process. Hidden within agent traces are rich records of how to make these systems better: what users ask for, when agents stumble, and how agents interact with the external world. Judgment is the most performant monitoring and improvement layer for long-horizon agents. We’re excited to see what you build!

    #Product updates#Agents#English text is being prepared.

  16. Testing CatalogProduct release
    AI score64

    Sentient Launches Open-Source On-Device AI Model Sentient OS

    Sentient has released the open-source AI model Sentient OS, which can collect context from local files and messages and execute tasks on the user's computer. Its Sidekick feature allows users to hand off tasks such as open forms or emails on the screen for it to complete using its own cursor. What it learns is stored locally as plain-text files that users can open and edit; the first 1,000 users to enter FINAL_1000 in the app can use it free for life.

    Quoted postAditya Vellanki · @aditya_vellanki

    Today, we're launching Sentient OS (YC F26), on-device AI that understands your life overnight with a local LLM. This unlimited inference enables 2 magical experiences 🧵 https://t.co/SFFAAnrxdj

    Same story · 2 reports

    #Product updates#Open-source ecosystem#Agents

  17. Testing Catalog
    AI score40

    Microsoft MAI Cyber 1.1 Flash launches on MDASH

    Microsoft announced that MAI Cyber 1.1 Flash is now live on MDASH. According to the official statement, its performance in the internal MDASH vulnerability scanning evaluation is comparable to MAI Cyber 1 Flash, with the cost per scan reduced by 18%. The post cited a statement from Mustafa Suleyman: over the past 30 days, user adoption of the MAI Cyber model among security tools doubled to 50%, with token efficiency improved by 20% without sacrificing quality.

    Quoted postMustafa Suleyman · @mustafasuleyman

    Our cyber models are getting really strong. MAI Cyber model in our security harness has doubled user adoption to 50% in the last 30 days. We also made it 20% more token efficient without sacrificing any quality. https://t.co/bslN1MTO1Z

    Same story · 2 reports

    #Product updates#Safety / alignment#Microsoft

  18. Mustafa SuleymanModel releases
    AI score58

    Microsoft MAI-Image-2.7-Preview Rises to Second on the Image Editing Leaderboard

    Mustafa Suleyman announced that MAI-Image has become the world's second-ranked image editing model, saying it once again surpasses Google, Meta, xAI and GPT-Image-2.5-Flare, and that it is available to try in MAI Playground. The Artificial Analysis evaluation he cited shows that Microsoft MAI-Image-2.7-Preview ranks second on AA-Image-Editing v2.1, behind only OpenAI's GPT Image 2.5 Sunburst (max), scoring higher than MAI-Image-2.6 on all 7 editing actions, including gains of +57 Elo in composition and framing, +49 in text or symbol editing, and +48 in scene and style editing, and ranking first on 4 of the actions. The model currently offers image editing only in MAI Playground, with text-to-image functionality said to be coming soon.

    Quoted postArtificial Analysis · @ArtificialAnlys

    Microsoft's MAI-Image-2.7-Preview takes #2 on AA-Image-Editing v2.1, close behind OpenAI's GPT Image 2.5 Sunburst (max) and a clear step up from MAI-Image-2.6 MAI-Image-2.7-Preview is Microsoft AI's newest image editing model, succeeding MAI-Image-2.6, released in August. In our testing, MAI-Image-2.7-Preview scores higher than MAI-Image-2.6 on all 7 editing actions we measure, and ranks #1 on 4 of them: Scene & Style Edit, Text or Symbol Edits, Reasoning-Based Edit, and Enhancement & Restoration. This new model improves on MAI-Image-2.6 with the biggest gains in Composition & Framing (+57 Elo), Text or Symbol Edits (+49) and Scene & Style Edit (+48). We see its best use cases to be in Live-Action Film, Consumer and Frontier edits, where it ranks #1. MAI-Image-2.7-Preview is available today in MAI Playground for image editing only, with text to image stated to be coming soon. Congratulations to @MicrosoftAI on the release! See below for our detailed analysis and example outputs of MAI-Image-2.7-Preview in the Artificial Analysis Image Arena 🧵

    Same story · 7 reports

    #Evaluations / benchmarks#Image generation#Microsoft

  20. Dan ShipperOpinion
    AI score71

    Dan Shipper shares an interpretation of an OpenAI internal model challenging ETH

    Dan Shipper shared Rishabh Karnawat's interpretation of an OpenAI internal model's claim to have broken the Exponential Time Hypothesis (ETH), calling it an excellent explanation. According to the interpretation, the model found an algorithm for solving 3-SAT faster than ETH permits, and the proof was checked in Lean; Impagliazzo and Paturi, who proposed ETH, along with MIT's Ryan Williams, are leading a public review. If the refutation holds, many complexity conclusions that rest on ETH as a premise would lose their foundation, but the new algorithm's absolute speed is still slow, and it does not amount to solving P versus NP.

    Quoted postRishabh Karnawat · @rishabhkio

    Many of you don't understand how serious this is, so let me break it down. In computer science there is a famous kind of puzzle called 3-SAT. You are given a long list of yes-or-no switches and a set of rules about how they must be arranged, and you have to find a combination that satisfies every rule. The obvious way to solve it is to try every combination, but each new switch doubles the number of possibilities, so a puzzle with a few hundred switches would take longer than the age of the universe to brute force. For decades, the smartest people in the field found clever tricks that made this a little faster, but every trick still hit the same wall: the time required kept exploding as the puzzle grew. About 25 years ago, two computer scientists, Russell Impagliazzo and Ramamohan Paturi, working with Francis Zane, turned that wall into a formal idea. They proposed the Exponential Time Hypothesis, which says this explosion is not a failure of human cleverness but a law of nature. No algorithm, however brilliant, will ever solve 3-SAT without the time blowing up exponentially. Nobody could prove it, but almost everyone believed it, and it became one of the foundations of modern computer science. Researchers used it as a measuring stick. If you could show that solving some other problem quickly would also crack 3-SAT quickly, you had shown that problem was hopelessly hard too. Hundreds of papers, across scheduling, graphs, logic and much more, rest on this one assumption. Now an internal AI model at OpenAI claims to have broken it. It says it found an algorithm that solves these puzzles meaningfully faster than the hypothesis allows, and it has checked the proof in a computer verification system called Lean. And here is the remarkable part: the very people who proposed the hypothesis, Impagliazzo and Paturi, along with MIT's Ryan Williams, are now leading the public review of whether a machine has disproved their own idea. So what happens next? If the disproof holds, a large part of theoretical computer science has to be revisited. Every result that said "this problem cannot be solved faster, assuming ETH" loses its foundation, and researchers will have to work out which of those walls were real and which were simply where we stopped looking. It does not mean everything becomes easy. The new algorithm is still slow in absolute terms and this is not the famous P versus NP problem being solved. But it means our map of what is possible was wrong in a place we were very confident about, and it was an AI that redrew it. If the disproof fails, the hypothesis survives and the field breathes out, but we will still have watched an AI produce serious, checkable work on one of the deepest questions in the discipline, closely enough that the founders of the idea had to take it seriously. Either way, something has shifted.

    #English text is being prepared.#English text is being prepared.#OpenAI