All AI updates
ITHome · AISame story · 2 outlets ITHome · AI World Labs founder Fei-Fei Li and AMD CEO Su Tseng Feng had their first joint appearance: robotics will become the 'new engine for computing power', with the market size potentially reaching tens of billions of US dollars in the coming years.
World Labs founder Fei-Fei Li and AMD CEO Su Tseng Feng had their first joint appearance at the CoRL 2026 conference, where Su Tseng Feng mentioned that the market size for physical AI, especially in robotics, could reach tens of billions of US dollars in the coming years. Fei-Fei Li believes that robotics will drive computing power into the next stage of development and emphasized that AI is computing and computing is AI.
ITHome · AI Honor 700 series is fully equipped with a 9000mAh Qinghai Lake battery
The Honor 700 series will be fully equipped with a 9000mAh Qinghai Lake battery, supporting 80W wired fast charging and 50W wireless fast charging, focusing on ultra-long battery life. The series adopts floating glass craftsmanship and an AI dynamic back screen. The back screen features a 200 million pixel selfie, AI interaction, and heart rate and blood oxygen detection functions without increasing the overall thickness. The standard version is equipped with the Snapdragon 8s Gen4, and the Pro version is equipped with the Snapdragon 8 Gen5.
ITHome · AI6565 Hua Hong Information is expected to mass-produce next-generation CPUs with performance comparable to international standards.
Haihong Information is expected to mass produce a new generation of CPUs next year, with performance comparable to international standards. The company's product layout includes DCUs supporting large model training, CPUs for AI data centers, and edge computing CPUs integrated with CPUs, GPUs, and NPUs.
ITHome · AISame story · 2 outlets Runway launches robot model Praxis-1, 15.6 times faster and cost reduced to 1/51
Runway has launched the robot model Praxis-1, based on the Runway video foundation model with 27 billion parameters. In the RoboLab-120 benchmark test, it achieved a success rate of approximately 46.2%, with a speed increase of 15.6 times and a cost reduction to 1/51. This model is licensed under Apache 2.0 and both weights and code are open-sourced.
Same story · 2 outletsITHome · AI6565 The 2027 version of the Aurora Alpha S5 will be launched on November 16: pre-sales start at 12.18 ten thousand yuan, standard configuration includes LiDAR.
The 2027 version of the ARCFOX Alpha S5 went on sale on November 16, with a pre-sale price starting at 121,800 yuan, equipped with a 192-line LiDAR, and featuring the Beiqi Yuanjing intelligent driving assistance system, supporting NOA navigation assistance for urban and expressway areas.
ITHome · AI Apex Compute releases ANR Cross-Platform AI Rendering Technology test version, challenging NVIDIA DLSS
Apex Compute has released the ANR cross-platform AI rendering technology test version, which includes AI frame generation, super-resolution, and ray reconstruction features, compatible with NVIDIA, AMD, Intel graphics cards, and Steam Deck. Currently, only the AI frame generation feature is open for testing.
ITHome · AI6565 Microsoft allows users to turn off Bing AI answers, but installing an extension requires setting Bing as the default search engine
Microsoft has launched the browser extension Microsoft Bing AI Search Choice, which allows users to turn off AI-generated answers in Bing Search. However, it forces Bing to be set as the default search engine during installation; users can also get traditional search results without AI answers by adding the '-ai' operator.
36Kr · AISame story · 2 outlets6565
OpenRouter BlogHow OpenRouter Builds Self-Correcting Localization Systems
OpenRouter currently supports 8 languages including Simplified Chinese, Spanish, Japanese, and Korean, and localization features have entered Beta testing. Its system achieves self-correction by detecting changes in strings in the code daily, translating them using GPT-6.1 Sol, and then verifying through code checks and a second model. When translation fails, the system retries with the reason, recording errors as rules or test cases. The entire process runs within OpenRouter, with translation data stored in the code-side Git repository.
36Kr · AI6565 Zhang Yiming on ByteDance AI: as early as 2029 or 2030 to catch up with OpenAI and Anthropic
Zhang Yiming recently stated at a ByteDance AI all-hands meeting that ByteDance has the capability to compete with OpenAI and Anthropic, but catching up to the cutting edge technology may not be until 2029 or 2030. The most pressing issue for ByteDance's Seed series models currently is that the scale of their pre-training dataset is still smaller than that of top-tier AI companies overseas.
36Kr · AISame story · 2 outlets6565 SpaceX is developing the Kessel AI chip rack to reduce reliance on NVIDIA
SpaceX is developing the Kessel AI chip rack, using a passive PCB middle board to replace the NVLink cable box, retaining 72 GPU modules, 18 computing trays, and 9 switching trays, planning to procure chip modules directly from NVIDIA and purchase the remaining components independently to bypass NVIDIA's L6 computing board and enhance supplier selection and procurement autonomy.
36Kr · AI6565 Anthropic: AI is expected to conquer difficult diseases like cancer within 10 years
Anthropic CEO Dario Amodei predicts that AI will be able to conquer difficult diseases such as cancer and Alzheimer's in 10 years. Anthropic has recently expanded its drug development business by acquiring several biotechnology startups and has launched internal new drug development efforts.
Hugging Face Daily Papers
TOME-Math: A mathematical proof supervisory corpus based on structured textbooks
TOME-Math transforms 2,118 university mathematics textbooks into approximately 380K structured proof supervision records, preserving chapter order, proposition-proof links, and dependency references, used for fine-tuning to improve the mathematical reasoning ability of language models by 16.7 percentage points, surpassing all three public instruction datasets. This corpus will be released to promote structured mathematical supervision research.
ITHome · AISame story · 2 outlets6565 Tongxin Intelligent completes billion Yuan financing, plans to conduct prototype verification for its first AGI chip in December
Tongxin Intelligence has completed a multi-million dollar seed round and angel round of funding, with investors including Tsinghua University Alumni Seed Fund; its first reasoning chip supporting AGI algorithms plans to conduct FPGA prototype verification in December.
ITHome · AI6565 Microsoft's built-in photo app for Windows 11 will integrate Adobe AI editing features.
Microsoft plans to test the new Windows 11 photo app with a preview version in December, integrating AI editing features from Adobe Express to enable personalized template application and streamline the Adobe Express design process.
Hacker News · 300+3737 NVIDIA CEO Huang Renxun is reportedly involved with a senior executive who reports to him
Reports suggest that Huang Renxun has a relationship with senior executives who report to him, which has caused attention. The news comes from a hot discussion post on Hacker News.
36Kr · AI6565 SYNAPUFF has secured a multi-million seed funding round to create an interactive intelligence base for AI Native terminals.
SYNAPUFF has completed a multi-million seed funding round to develop an interactive intelligence base for AI Native terminals, enabling agents to understand context, judge interaction boundaries, and dynamically adjust behavior. The company has proposed the 'P Framework,' which uses an interaction loop of 'trial—receive response—decision,' combined with dynamic interaction models, personality strategies, and non-textual language action generation models, to form an independent interaction decision framework adaptable to different terminals.
MIT News · AIMIT researchers have developed new technology to accelerate and enhance the credibility of AI models for climate prediction.
MIT researchers have developed a training technique that significantly reduces data volume and improves the performance and adaptability of climate simulators (AI models), enabling them to quickly simulate the impact of human activities on future climate in seconds, such as predicting temperature or sea level changes, thereby more reliably testing thousands of climate scenarios and identifying unintended consequences of regional interventions. The algorithm generates optimized synthetic climate scenario data instead of relying on future emission scenarios based on real-world trajectories, significantly shortening production time and allowing the model to accurately predict scenarios not present in the training data.
Latent SpaceHow Jev from TypeSafe AI is Leading the System One Model Revolution
TypeSafe AI's Jev is a non-generative system-one model that can return structured choices, scores, or probabilities, faster and cheaper than inferential LLM (system-two models). Jev is suitable for zero-shot flexibility, requiring no repeated fine-tuning, and users can achieve reliable intelligent automation without needing to understand its internal model type. Jev can be integrated into code, enabling the construction of smarter software through rapid calls, with its application potential likened to the exploration stage of the early internet.
Latent SpaceSame story · 7 outletsSame story · 7 outlets- Google AI Blog · Google launches Gemini Decisions API and Gemini Decision 1 modelFirst
- The Decoder · Google launches Gemini Decision 1 and Gemini Decisions API, capable of processing 1000 issues at a time
- MarkTechPost · Google has released the Gemini Decisions API and Gemini Decision 1, priced at $0.10 per million tokens.
- Simon Willison · Google releases Gemini Decisions API
36Kr · AISame story · 2 outlets6565 Elon Musk responds to Anthropic or stopping the use of SpaceX computing power: The company is happy to reclaim it
Elon Musk responded to the possibility that Anthropic may stop renting SpaceX computing power, stating that the company would be happy to take it back. Musk believes the actual situation may be the opposite.
Same story · 2 outletsHugging Face Daily Papers6565
One Table Benchmark launched: a benchmark test for general table models in generation, interpolation, and prediction tasks
The One Table Benchmark tested the table generation, imputation, and prediction capabilities of 46 models on 56 datasets, finding that these tasks share conditional information but have different fidelity and target interactions. Fixed models exhibit a trade-off between utility and fidelity, while frozen base models perform outstandingly in prediction and numerical imputation, indicating that unified table models have the potential to improve classification reconstruction and joint sampling.
LangChain BlogHow to build consistent, high-quality pre-sales materials with LangChain GTM Agent
LangChain's internal GTM Agent generated 1540 pre-sales materials for sales representatives within 12 weeks, improving sales efficiency by 47% (the probability of material transactions moving to the next stage increased from 16% to 47%), saving an average of 3.37 hours of human labor cost per material.
ITHome · AISame story · 2 outlets6565 36Kr · AISame story · 2 outlets6565 Baidu founder Li Yannhong: Reasoning will be the main driving force for the long-term sustained growth of AI chips.
Baidu founder Robin Li stated at Baidu World 2026 that reasoning will be the primary driver of long-term sustained growth for AI chips. Current AI computing power demand has shifted from training-centric to a balance between training and reasoning, with the large-scale deployment of intelligent agents driving a surge in reasoning computing demand. Baidu's Kunlun Chip M100 is now on the market, with the M300 expected to be launched next year.
ITHome · AI6565 Guo Mingzhi talks about the Huang Renxun cheating scandal: NVIDIA might need to start planning for a successor.
Kuo Ming-chun believes Huang Renxun is crucial to NVIDIA, capable of transforming technology into products that the company invests in resources for, and driving supply chain investments. Huang Renxun seems unlikely to retire easily, and NVIDIA may need to reduce its decision-making reliance and plan for a successor.
ITHome · AI Baidu Tianchi 1024 Card Super Node Released: Can support training of 5 trillion parameter models, to be launched next year.
Baidu Tianchi's 1024 card super-node is launched, equipped with optical engine technology, inter-card interconnection bandwidth increased 3 times, training performance increased 2.5 times, and inference performance increased 3 times. Accompanied by G3.5 KV Cache GPU memory acceleration cabinet, long context inference performance increased 2 times, multi-round dialogue cache hit scenario TTFT decreased by 90%, and throughput increased 7 times. A single node can support the training of large models with 500 billion parameters, with the ability to scale to ten thousand cards, and will be launched next year.
ITHome · AISame story · 2 outlets Baidu's Kunlun Core M300 will be launched next year, and multiple chips based on the new N architecture will be released in 2028.
Baidu announced that its flagship AI chip, the Kunlun Core M300, will be launched next year, focusing on comprehensive performance upgrades; by 2028, multiple chips based on the new N architecture will also enter the market, precisely matching differentiated application scenarios.
ITHome · AI6565 CNET released a battery drain test of 32 phones: Apple iPhone 18 Pro Max has the best overall endurance performance
CNET tested 32 phones and found that the Apple iPhone 18 Pro Max lost 1% battery in a 45-minute stress test and consumed 6% battery in a 3-hour video test, showing the best overall battery life. iPhone Duo topped the stress test, and Pixel 11 Pro XL ranked third. The test accumulated over 120 hours, and the iPhone 18 Pro Max's victory was due to its larger battery, iOS 27 optimization, and improved energy efficiency of the A20 Pro chip.
ITHome · AI6565 Tencent denies 50% layoff rumors: claims it is a misinterpretation of organizational restructuring and personnel adjustments
Tencent denied the 50% layoff rumors, stating it was a misunderstanding of organizational restructuring and personnel adjustments; insiders revealed that Tencent's 28th anniversary internal meeting proposed proactive changes for the AI era, encouraging 50% of employees to focus on core business efficiency improvement, while the other 50% explore AI Native new spaces, not layoffs.
36Kr · AISame story · 2 outlets4545 Doubao Work has integrated models such as Kimi 3.1 Pro, GLM 5.3, and DeepSeek V4.1 Flash, and is offering limited-time promotions.
Dou Bao has officially integrated multiple third-party models such as Kimi 3.1 Pro, GLM 5.3, and DeepSeek V4.1 Flash, and has launched a limited-time special offer plan. Users can select the corresponding model according to their actual work needs to complete tasks.












