Skip to content
Live
#Other
Nov 11Wed
  1. Hugging Face Daily Papers65

    Longtail-SWE: High-quality long-tail software engineering task set

    Longtail-SWE has released 202 problem-solving tasks covering 13 underrepresented programming languages, for engineering problems that are scarce in existing benchmarks. Built through a solver-judge-repairer pipeline, the model was evaluated under 9 configurations, including state-of-the-art models, analyzing task-solving performance, consistency of repeated attempts, and resource usage. All benchmark instances, execution environments, tools, and agent trajectories have been open-sourced.

  2. The Verge · AI65

    Anthropic has released the AI office suite Claude Meetings, Sheets, and Whiteboards.

    Anthropic has launched an AI office suite for non-technical knowledge workers, including Claude Meetings, Claude Sheets, and Claude Whiteboards, supporting features like meeting notes, shared whiteboards, spreadsheet editing, and presentation video generation, all currently in Beta testing. Some tools are available to paid plans, and Claude Data has been officially opened to all paid users.

Nov 10Tue
  4. ITHome · AI65

    Walt Disney tests AI parking robots

    Walt Disney collaborates with Field AI to test AI parking robots, helping visitors find parking spots in the park; the robots draw design inspiration from 'Cars,' equipped with screens and mechanical arms to provide guidance, but Disney has not announced deployment time.

  5. ITHome · AI40

    Honor 700 series preview: debuts back screen arbitrary stick function, Ju Jingyi confirmed as spokesperson

    The Honor 700 series is previewing the back screen arbitrary stick function, supporting expressions, GIFs, etc., which can be pressed and dragged onto the back screen, and also supports rear selfie, AI interaction, heart rate and blood oxygen detection. The series will be equipped with a back screen in full, with a body thickness of 7.96mm, internal code Flora, aims to inherit the flagship imaging capabilities and explore new AI play methods, the spokesperson is temporarily set as Ju Jingyi.

  6. Hugging Face Daily Papers

    Fast and Exact Relative-Position Attention via Structure-Aware Kernels

    A precise tile attention algorithm integrates additive relative position terms into the computation similar to FlashAttention, reducing memory and computational overhead, accelerating long-context relative position Transformers. The method (1) uses offset-aware tiles to avoid redundant index and memory access, (2) only computes necessary key/query relation products and reuses them in forward and backward passes, and (3) uses offset-based gradient sharding and aggregation to eliminate atomic contention.

  7. ITHome · AI65

    The GAC Aion Ray 7 will be officially launched on November 12: Equipped with Huawei's 192-line LiDAR, powered by the Weiyuan Intelligent L4-level Robotaxi-derived large model

    The GAC Aion Ray 7 will be launched on November 12, equipped with Huawei's 192-line LiDAR, and powered by the Weiyuan Intelligent L4-level Robotaxi-derived large model. The car is nearly 5 meters long, with a drag coefficient of 0.22Cd, and a CLTC electricity consumption of 9.7 degrees per 100 kilometers, offering a CLTC pure electric range of 700km.

  8. ITHome · AI

    Anthropic has launched the Claude Fable 5.2 and Claude Mythos 5.2 AI models, with a 20% reduction in API call fees.

    Anthropic has launched the Claude Fable 5.2 and Claude Mythos 5.2 models, the latter targeting professional institutions. Fable 5.2 has input and output costs per million tokens of $8 and $40, respectively, a 20% reduction from 5.1. Generation speed has been increased by up to 30%, with an optional fast mode that runs 2.5 times faster but at a higher price.

  10. The Verge · AI65

    Disney and Adobe jointly develop proprietary generative AI models

    Disney and Adobe announce collaboration to develop proprietary 'artist-led' generative AI model, supporting Disney's production process. The collaboration will give Disney early access to Adobe's latest technology for content creation in theme parks, cruise ships, ESPN graphics, etc. Both parties will use Adobe Firefly Foundry to explore AI models that fit Disney's unique knowledge system (character attributes, environment, visual style), accelerate work, and develop more creative projects.

    Same story · 2 outlets
  11. The Decoder65

    Disney collaborates with Adobe, using Firefly to create AI models that meet brand safety standards

    Disney and Adobe announce a multi-year collaboration to create custom AI tools for the entertainment empire. Disney will get early access to Adobe's new technologies for theme parks, cruise ships, and ESPN content. Adobe will build brand-specific models for Disney through Firefly Foundry, which will be trained based on Disney characters, environments, and visual styles to generate commercially safe, production-ready outputs.

    Same story · 2 outlets
Nov 9Mon
  4. ITHome · AI65

    The Xiaomi 18 Pro series will be launched internationally on November 17 at 14:00 Paris time.

    The Xiaomi 18 Pro series will be launched internationally for the first time in Paris on November 17 at 14:00 local time, starting at a price of 1,549 euros. The series was released in China on September 23, featuring the Snapdragon 8 Ultra, a 200 million pixel main camera by Leica, and an AI adaptive back screen. The global version is available in black, white, and burgundy (exclusive to the global market).

  5. ITHome · AI65

    An engineer has developed an open-source script, Ram-Altman, which reports daily the amount of idle RAM to OpenAI CEO Altman.

    Engineer IainKay developed the open-source script Ram-Altman, which monitors computer memory usage through the Linux system. When available memory exceeds 10% of the total capacity, it automatically sends an email to OpenAI CEO Sam Altman daily to report idle RAM amounts, and has prepared various email subjects, body texts, and signatures. The project has been uploaded to GitHub for free use, but the developer warns users to avoid mass emailing to bother Altman.

  7. ITHome · AI65

    Japanese AI unicorn Sakana AI is developing its first mobile app

    Sakana AI is developing its first mobile application 'Sakana AI App' and is currently recruiting beta test users. The company was jointly founded by former Google and Stability AI executives and the author of the Transformer architecture paper. Using the 'evolutionary model merging' technology, it has achieved a valuation exceeding 1 billion US dollars in less than a year.

  8. ITHome · AI35

    Cao Cao Mobility, in collaboration with Ant Group, has established AI driver home posts in Hangzhou, Xiamen, and Chongqing.

    Cao Cao Mobility, in collaboration with Ant Group, has rolled out AI Health Posts at driver homes in Hangzhou, Xiamen, and Chongqing in batches. The Hangzhou and Xiamen stations opened on November 4 and 6, respectively, with the Chongqing station scheduled to launch on November 12. The posts offer free health services such as blood pressure and blood sugar measurements, and will be gradually expanded to more operational centers in the future.

  9. Hacker News · 300+34

    BenchUp – Open source SketchUp alternative built with Rust and WebGPU

    BenchUp is a free and open-source 3D modeling application for woodworking and furniture design, available on Windows, macOS, Linux, or desktop browsers. Users can accurately measure and plan shelves, tables, cabinets, or workbenches, support drawing lines, rectangles, circles, and freehand outlines, create shapes through Push/Pull tools, and add dimensions and grouping. The project is developed based on Rust and WebGPU, using the MIT license, and is free to use without an account.

  11. ITHome · AI60

    LG AI Research Institute announces the free opening of the K-EXAONE 2.5 model API to the public

    LG AI Research Institute has opened the K-EXAONE 2.5 model API for free, allowing any user to use it at no cost for development and research. This self-developed model enhances the capabilities of the intelligent agent through additional training and performs equally well as DeepSeek-V4.1-Flash and GLM-5.3 in benchmark tests such as AutomationBench and SWE-Bench Pro.

  12. ITHome · AI38

    Apple plans to launch an autonomous Siri with an agent next year

    Apple plans to launch an autonomous and agent-based Siri as early as next year, aiming to shift from responding to commands to proactively managing tasks and representing users to complete transactions. According to Bloomberg reporter Mark Gurman, the new CEO John Tuennies has urged the software engineering chief to accelerate the development of the AI business, dissatisfied with only completing the necessary feature upgrades for Siri.

Nov 8Sun
  1. The Decoder52

    Turing Award winner Richard Sutton has 'bitter lessons' about scaling large models

    Turing Award winner Richard Sutton states that raw computing power is insufficient to achieve intelligence, AI science needs new methods and is far from completion. He criticizes the limitations of simply scaling large language models (LLMs), calling synthetic data a 'big mistake' and human expertise the bottleneck for expansion. The industry has sparked widespread discussion, with some researchers agreeing that new learning algorithms are needed, but others point out that the algorithms themselves are crucial for scaling.

  4. Hacker News · 300+65

    Farewell Claude, hello Pi: A veteran Claude Code user's experience and reflection on migrating to Pi

    A long-time Claude Code user decided to migrate to Pi due to persistent minor frustrations. Claude Code lacked stability and quality control during rapid feature iterations, such as issues with the AGENTS.md file, VSCode integration limitations, and frequent changes in system prompts, ultimately leading the user to switch to Pi. Pi offers stability, minimalist design, and high customization, with the user already implementing multi-tab text selection functionality in 20 minutes by writing an extension.

  5. ITHome · AI42

    Shinwasei Kogyo Qiyuan Robot has a monthly active rate of over 98%, with user interactions exceeding 100 times per day.

    Shinwasei Kogyo Qiyuan Robot has a monthly active rate of over 98%, with an average of over 100 interactions per user per day, and an average online time of 3 hours per day. Qiyuan and T1 have completed their first OTA upgrades. Q1 has added 6 major official characters and English speaking practice, and T1 has added terrain recognition and remote control functions.

  6. ITHome · AI55

    Xianfeng launches the 'Mijia AI Smart Graphene Oil radiator': Connects to the Mijia App, 2500W power, 499 yuan

    Xianfeng lists the 'Mijia AI Smart Graphene Oil radiator' on Youpin, connecting to the Mijia App, equipped with a 2500W rated power and 17 patented heat wave plates, increasing the whole house heating speed by 19%, priced at 499 yuan. The product supports temperature adjustment from 15-35°C, 24-hour timer on/off, and includes an enlarged drying rack and a separate humidifying box.

  7. MarkTechPost58

    Nous Research releases Arete-27B, reducing the promotional content of Qwen3.8-27B from 22% to 0.9%

    Nous Research has released Arete-27B, a 27B parameter model with open-weight quantization, which removes promotional content about China from Qwen3.8-27B through supervised fine-tuning. In 803 prompt evaluations, the promotional content ratio decreased from 22% to 0.9%, while retaining the agent skills of the Qwen class and being free from political bias.

  8. ITHome · AI65

    Sony launched the 'dual-path visual information fusion in-vehicle passenger dizziness detection' technology

    Sony globally launched the 'dual-path visual information fusion in-vehicle passenger dizziness detection' technology at the CIIE, which predicts passenger dizziness levels by analyzing the front road view and rear seat screen content. The evaluation results can be used to remind drivers to reduce hard braking or turning, or to recommend visually less changing content for content platforms. This solution reuses the ADAS front view camera and does not require a new rear seat camera.

  10. The Decoder65

    The thinking machine Inkling 2 Small used 316 profanities in benchmark tests.

    Vals AI researchers analyzed the thinking logs of Inkling 2 Small over 2,655 runs on 13 benchmarks and found 50 instances using 'fuck' or its variants, totaling 316 times, with 241 of them coming from two out-of-control runs. In one task, the model generated a single response of 500,000 characters while calculating asteroid trajectories, repeating the same sentence and including 158 instances of profanity. ProgramBench was the most affected, with profanity appearing in 20 out of 200 runs.