NAll news

All news

Saturday482 items

  1. MacRumors Apple Intelligence
    AI score26

    The MacRumors Show: Discussion of OLED MacBook Pro, iPad mini 8, and Other New Products

    On this episode of The MacRumors Show, we discuss Apple's newly announced OLED MacBook Pro, the iPad mini 8 with the A20 Pro, as well as the M6 iMac and the entry-level 14-inch MacBook Pro. The OLED MacBook Pro replaces the notch with Dynamic Island; the 14-inch model weighs 3 pounds and the 16-inch model weighs 4 pounds; it continues to use the M5 Pro and M5 Max, with maximum storage doubled to 16TB, and starting prices of $2,899 and $3,399. The iPad mini 8 switches to an 8.5-inch OLED display with a peak HDR brightness of 1,200 nits, becoming the first iPad with IP67 protection, with a starting price of $699.

    #English text is being prepared.

  2. Baseten Engineering BlogFirst-party
    AI score23

    Baseten launches autoscaling schedules: pre-warm replicas before traffic arrives to reduce cold start latency

    Baseten has launched autoscaling schedules, which can raise the minimum replica count in advance before predictable traffic arrives, so requests do not have to wait for cold starts. Each environment can be configured with up to 10 schedules, supporting daily, hourly, or one-time time windows; windows cannot overlap and must last at least 5 minutes, all schedules share a single time zone, and they can be configured via the dashboard, CLI, or Management API. When a schedule is not active, the environment uses its default settings.

    Autoscaling schedules

    #Product updates#English text is being prepared.

  3. 404 Media
    AI score44

    Proton CEO Andy Yen Cancels AI "Re-Interview"

    Proton CEO Andy Yen announced at an all-hands meeting on Friday that the company was canceling the planned AI "re-interview" that would have covered nearly the entire company, following an investigative report published by 404 Media. He acknowledged a communication failure but said he would continue to push AI, and stated that the related assessment metrics would be folded into performance reviews. Proton users had previously criticized the company on social media for downplaying security, and some called for Yen to resign.

    #Industry news#Safety / alignment

  4. QbitAI
    AI score20

    Utopai Studios Announces 2027 Slate: "The Most Serious Fart," "Cortés," and "Half Moon" to Hit the Big Screen

    The $1 billion-valued AI film and television unicorn Utopai Studios has announced its 2027 slate. "The Most Serious Fart," "Cortés," and "Half Moon" will successively arrive on the big screen, and the market will for the first time use its works to test the company's system for rebuilding the film industry's infrastructure. The company says it has completed the hardest first half of the journey—from technology validation to commercial validation to IP validation.

    #Industry news#Video

  5. QbitAIProduct release
    AI score65

    Baidu APP Releases Four Major AI Upgrades, Search AI API as Low as 4.9 Yuan per Thousand Calls

    At its product upgrade briefing on October 30, Baidu APP released four upgrades covering search, AI camera, GUI visualization, and AI interactive applications, and upgraded the Search AI API, with a minimum price of 4.9 yuan per thousand calls during the promotional period. Its search capabilities now cover more than 170 tool scenarios and can perform multi-step lookup and organization and deliver results. Baidu says its AI-native four-layer architecture drives recursive self-evolution of the model through a "retrieval—feedback—retraining" closed loop, improving overall iteration efficiency by more than 80%, and the application layer supports full-solution access via API, MCP, and Skill.

    #Product updates#Search#Agents

  6. QbitAI
    AI score22

    Lenovo Tianxi AI 4.5 upgrades across all devices, with "Super Mode" creating a super workplace partner

    Lenovo Tianxi AI 4.5 has completed a full-device upgrade and launched "Super Mode," using the Tianxi AI "unified core" to connect the AIPC, AI tablet, AI phone, AIoT, and AI host ecosystems. This upgrade strengthens contextualized hybrid AI experiences under the device-edge-cloud architecture, breaks down barriers between applications and data, and enables seamless flow of the office experience.

    Same story · 2 reports

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

  7. QbitAIProduct release
    AI score74

    ModelBest ForgeKernel Surpasses DeepSeek FlashMLA by 9.8% in Live Operator-Writing Session and Goes Open Source

    ModelBest's operator generation system ForgeKernel generated DeepSeek-V4.1's FlashMLA Prefill operator from scratch on an NVIDIA B200, running 9.8% faster than the human hand-written version, through 42 rounds of iteration over 22.5 hours and 22 GPU hours, with humans writing not a single line of code. After the live stream ended, the team, together with the OpenBMB community, announced the open-sourcing, and it topped three Ascend CANNBench leaderboards: 950DT, 950PR, and A3. The team and institutions including Tsinghua University published the JustGoal paper on arXiv, stating that when given only optimization goals, the model beat human-maintained open-source libraries on 8 of 10 operators on the B200.

    #Product updates#Agents#Open-source ecosystem

  8. QbitAIProduct release
    AI score55

    Zhiyuan releases GO-3 Preview and WITA-OMNI dual models, and proposes the Worken work-hour unit

    At its Innovation Day, Zhiyuan released the embodied foundation model GO-3 Preview and the omni-modal interaction large model WITA-OMNI; the former topped three benchmarks—RoboDojo, RoboTwin, and RoboColiseum—while the latter ranked first on four benchmarks: Daily-Omni, TimeLens-Bench, EmbSpatialBench, and MMAU. It also launched the third-generation hardware architecture AHA3, the heavy-duty robot Jingling G2 Max, and the embodied native operating system Lingqu OS, and plans to fully open-source Lingqu OS 1.0 by the end of the year. Zhiyuan also proposed Worken, a unit for measuring a robot's effective intelligent work hours, with the formula being the number of active robots multiplied by autonomous working duration multiplied by the output efficiency ratio.

    Same story · 12 reports

    #Industry news#Embodied AI#Model releases

  9. Google BlogFirst-party
    AI score67

    Google Receives First AI Reply from Project Suncatcher Prototype Satellite's On-Orbit TPU

    Google received the first reply from the Project Suncatcher prototype satellite, with the on-orbit data-center-class TPU successfully running the open Gemma 4 model and returning a haiku about space. The satellite was launched on October 1, and the host and TPU tray have previously completed power-on and early machine learning tests. Next, Google will test how the TPU performs under orbital thermal stress and radiation exposure, paving the way for the 2027 two-satellite interconnection goal.

    Same story · 8 reports

    #Industry news#Google#Reasoning

  10. TechCrunch AI
    AI score59

    Google Suncatcher's first data center satellite goes online and completes in-orbit inference

    Google Suncatcher's prototype space data center satellite has gone online and communicated with the ground; earlier this month it launched aboard a SpaceX Falcon 9 rocket along with more than 100 other spacecraft. The satellite carries a single Google TPU with a maximum power generation of 1 kW, and due to power constraints it can only be powered on for about twenty minutes at a time; after about three weeks of in-orbit commissioning, ground engineers this week sent the first prompt, asking the pre-installed Google DeepMind Gemma 4 model to write a haiku about space and to report on the orbital conditions around the satellite. Subsequent tests will examine the chip's performance under thermal stress and the space radiation environment; if this TPU performs up to standard, Google plans to launch two more satellites carrying the chip in 2027 to validate higher power generation and inter-satellite optical links.

    Same story · 2 reports

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

  11. n8n BlogFirst-party
    AI score27

    What's the difference between rate limiting and throttling: n8n explains workflow backoff strategies

    n8n's post distinguishes API rate limiting from throttling: the former typically returns a 429 error when exceeded, while the latter delays or queues requests. When encountering a 429, you should first follow the Retry-After header in the response; otherwise, use exponential backoff to progressively lengthen the wait, set a maximum delay and retry limit, and add jitter to prevent multiple workflows from retrying simultaneously and creating new traffic spikes.

    #Tutorials / practice#English text is being prepared.#Agents

  12. AYi AI NotesOpinion
    AI score51

    The Full Story of the $6.3 Billion Binance Hot Wallet Hack: How Hackers Used EIP-7702 Delegation Authorization to Bypass Signatures

    AYi AI Notes used Claude Opus 5.5 to generate a retrospective video laying out the on-chain process by which $6.3 billion was moved out of Binance's hot wallet. The author argues that the key point is that the hackers did not steal private keys; instead, they exploited Ethereum's EIP-7702 delegation authorization to make Binance's signing system sign for the hackers' contract. The authorization contained no amount or recipient address, making it hard for traditional risk controls to detect. The article states that the hackers signed 50 authorizations in succession, with the 16th taking effect after matching the nonce, and then a single call moved out 167,000 ETH, 1.62 billion USDT and more than 60 other tokens, leaving only 1 ETH in the wallet.

    Quoted postAYi · @AYi_AInotes

    English text is being prepared.

    #Safety / alignment#OpenAI#Anthropic

  13. Karina Nguyen
    AI score28

    BallotBench adds a trading track

    Karina announced that BallotBench is adding a trading track. The cited Thoughtful post explains: they gave 9 models $50 each in real money on Kalshi plus $10K each in simulated funds, and had them place bets on the midterm elections based on their own forecasts; after 24 hours, Claude Fable was up $1.59 (+3.2%) and Gemini was up $0.80 (+1.6%).

    Quoted postThoughtful · @thoughtfullab

    Our AI forecasters now trade the midterms! We gave 9 models $50 each in real money on Kalshi + $10K each in simulation to place bets based on their forecasts. 24 hours in: Claude Fable is up $1.59 (+3.2%), Gemini $0.80 (+1.6%) Watch live: https://t.co/YgKV75W0fC https://t.co/vmtVan8ydc

    #Evaluations / benchmarks#Agents

  18. Fei-Fei Li
    AI score37

    Fei-Fei Li shares I3L robot imitation learning research

    Fei-Fei Li shared the I3L (Iterative Interactive Imitation Learning) research posted by Wensi Ai, calling it the team's latest work on closing the "last-mile" gap in long-horizon robotic tasks. The original post noted that long-horizon whole-body robot policies still fail during deployment.

    Quoted postWensi Ai · @wensi_ai

    Long-horizon, whole-body robot policies still fail during deployment. How do we best close this last-mile gap? 🤖 🚨 Introducing I3L: Iterative Interactive Imitation Learning for Whole-Body Mobile Manipulation Website: https://t.co/C58RmOS7Oq Paper: https://t.co/DZXnXLF1bI https://t.co/JbK5Dn5Jig

    #Papers/Research#Embodied AI

  19. Rohan PaulProduct release
    AI score51

    Subvocal launches glasses for silent mouthing input that can transcribe 200+ words per minute

    Subvocal has launched a pair of glasses that can convert silently mouthed words into text at a speed of over 200 words per minute. The principle is that radar passes from the frame through the face, and based on changes in the reflected signal as each word is silently mouthed, AI decodes it into text; at most, bystanders see only slight mouth movements. The product is now available for pre-order, and the team previously raised $1.3 million in funding from YC and Afore Capital.

    Quoted postAarush Yadav · @AarushYadav47

    We’ve raised $1.3M from YC & Afore Capital so you never have to type or speak again. Typing is slow. Talking into your computer is awkward. Subvocal fixes both Our glasses give you the speed of speech without you making a sound or touching a keyboard. It tracks your internal speech muscles with non-contact sensing and decodes silent, almost invisible mouthing into text at more than 200 words/min. Pre-orders are now live at: https://t.co/DzrwIpg3GR

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