Archive

UUnsloth21:26First-partyProduct release

AI score72

Unsloth releases GGUF version of DeepSeek-V4.1-Flash-1017

Unsloth releases a new GGUF version of DeepSeek-V4.1-Flash-1017, claiming it ranks highest in benchmark tests. It says Dynamic V3.1 leads on almost all benchmarks including GPQA Diamond, AIME25, LiveCodeBench v6, MMLU-Pro, and Aider, requires more than 64GB of memory to run, and includes links to the GGUF and the blog.

Quoted postDeepSeek · @deepseek_ai

🚀 DeepSeek-V4.1-Flash-1017 is here! Stronger benchmark performance, faster task completion, and shorter outputs than the previous version—a Pareto improvement in performance and cost. Now live on our website, app, and API. 🛠️ Optimized for @DeepSeekHarness, Tencent CodeBuddy, WorkBuddy, and more. Upgrade to DeepSeek Harness v0.2.1-rc.1 or later for the best experience with the new model. 1/4

Why read thisUnsloth has released a GGUF quantized version of DeepSeek-V4.1-Flash-1017, along with a 64GB memory requirement, making it easier for local deployers to evaluate.

#Model releases#Open-source ecosystem#DeepSeek

PPeter Steinberger21:26

AI score42

Peter Steinberger on Claude Code's new rule: you should build your own harness

A post claims that around October 27, Claude Code pushed down a new rule named "tengu_heron_brook" via Anthropic's servers, and it cannot be turned off: you can build original games, but you must not recreate a specific existing game. Commenting on this, Peter Steinberger said this is yet another reason to use your own harness rather than one whose permission scope a company can change at will.

Quoted postchasm · @chasmmmmmmmmmmm

ANTHROPIC IS CRACKING DOWN ON VIDEO GAME REWRITES 🚨 Claude Code picked up a new rule around Oct 27, sent from Anthropic’s servers under a flag called "tengu_heron_brook". There’s no setting to turn it off. The full rule: “Do build original games, games in a genre, and games from the user’s own designs and assets. Don’t build a recreation of a specific existing game, even if it is renamed or called original.” Is this the end of game rewrites, modding and everything in between?

#English text is being prepared.#Agents

EEthan Mollick21:26

AI score25

Ethan Mollick: The gap between labs' internal models and public models may be widening

Ethan Mollick points out that there is currently no effective way to measure the capability gap between a lab's strongest internal model and its strongest publicly released model. He speculates that this gap has continued to widen over the past year, especially after Fable was temporarily banned.

#English text is being prepared.#Safety / alignment

MMicrosoft Research21:22First-party

AI score35

Microsoft evaluates Copilot with synthetic enterprise data: from simulated tenants to real AI confidence

Microsoft built the Industry Evaluation Data Generation Platform, which uses fully synthetic enterprise environments generated from public data such as O*NET, job postings, and SEC filings to evaluate Microsoft Copilot, avoiding the use of any customer data. These simulated tenants contain realistic organizational structures, meeting and email histories, can cover issues missed by general benchmarks, and have formed Copilot's first standardized evaluation baseline.

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

TThe Verge AI21:19Opinion

AI score65

Altman Responds to Math Community Doubts: OpenAI's Math Results Releases Will Not Stop

Sam Altman said on the Cheeky Pint podcast that the public's right to understand how the universe works outweighs the impact of OpenAI's result releases on the math community, and that the company needs to improve its communication but will not stop releasing. He admitted that OpenAI's investment in math problems once got out of control, and said the company has hundreds of math discoveries beyond its published results that remain undisclosed; he also estimated that OpenAI has spent hundreds of millions of dollars on math problems, of which the Navier-Stokes solution cost about $10 million. The math community, meanwhile, criticized OpenAI's releases for lacking attribution, context, and verification, and also expressed dissatisfaction that details of results such as the Yang-Mills theory have not been disclosed.

Same story · 2 reports

Why read thisEnglish text is being prepared.

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

TThe Verge AI21:19

AI score66

Anthropic Files a 335-Page S-1, With 94 Pages of Risk Factors Filled With "Miracles" and "Existential Risk"

Anthropic filed a 335-page S-1 prospectus on Monday, including as many as 94 pages of risk factors; the founders' letter states that advanced AI could usher in an "age of miracles" but could also pose an existential risk to humanity. The prospectus also states that the company has a history of net losses, has taken on hundreds of billions of dollars in financial obligations to secure computing power, and will build its own data centers; the author notes that key information for assessing profitability, such as third-quarter results, is missing. Jay R. Ritter, a professor at the University of Florida, noted by comparison that Apple's 1980 IPO prospectus was only 47 pages and contained no risk factors.

Same story · 120 reports

Why read thisThe article breaks down the 94 pages of risk factors and the "anti-pitch" rhetoric in Anthropic's 335-page S-1, revealing the gap between an AI company's IPO narrative and its financial reality.

#Industry news#Anthropic#OpenAI

SSierra：Blog21:16First-party

AI score58

Sierra's annualized revenue surpasses $400 million, with over $150 million added in Q3 alone

Sierra announced that its annualized revenue (ARR) has surpassed $400 million, adding over $150 million in Q3 alone—more than its entire ARR for the previous year. The company said it took six months to go from $100 million to $200 million, and only five months to go from $200 million to $400 million; it currently serves nearly 50% of Fortune 50 enterprises. Sierra also mentioned that a major U.S. retailer with over $40 billion in revenue used Ghostwriter to go from first meeting to agent launch in 27 days, and that it is working with more than 50 companies including Meta, Shopify, Stripe, and Walmart to advance Poppy, an open standard for personal agent interaction.

Why read thisSierra disclosed its ARR growth trajectory from $100 million to $400 million and provided a specific case of a customer using Ghostwriter to quickly launch an agent.

#Industry news#Agents

4404 Media21:14

AI score22

Apple Enters the "Cheap and Crappy" Era

Apple's new foldable iPhone Duo developed screen cracking issues just days after launch, and the company subsequently announced a touchscreen laptop, which the author sees as relying on hardware gimmicks to stay relevant amid the AI shock. New CEO John Ternus previously led projects such as the Touch Bar and the butterfly keyboard, and in a Time interview he called other foldable designs "cheap and crappy." The author recounts, through personal experience, the reasons he switched to Android back in the iPhone 4 era.

#English text is being prepared.#Industry news

MMIT Tech Review21:14

AI score11

The Download: GLP-1 and eating disorders, self-aware cities, and forever chemicals

MIT Tech Review's The Download notes that a new report shows the global incidence of anorexia nervosa and bulimia nervosa rose 71% between 1990 and 2023, and clinicians have observed that GLP-1 weight-loss drugs can exacerbate the problem. The same issue also features Charlie Jane Anders's short science fiction story "Love is Infrastructure," as well as a report on progress in remediating soil and water contaminated by PFAS "forever chemicals."

#Industry news

TTechCrunch AI21:13

AI score50

Runware wants to mass-produce AI data centers like cars, targeting 30 megawatts per day by 2028

AI infrastructure company Runware plans to mass-produce its modular data center Sonic Inference Pod on an assembly line at a factory in Romania, targeting a daily production capacity of 30 megawatts by 2028, equivalent to nearly 1 gigawatt per month. Customers can bring their own GPUs or rent compute from Runware. The Pod's onboarding target is to complete installation and deployment within 48 hours, and construction time will be cut from three weeks to one day. Runware has raised more than $60 million in cumulative funding and says the Sonic Pod has received $1 billion in intent, corresponding to more than 20 megawatts of capacity.

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

EElevenLabs Blog21:13First-party

AI score34

How TELUS Digital Used ElevenAgents Outbound Calling Agents to Cut Early Churn by More Than Half

TELUS Digital built an outbound voice agent based on ElevenAgents that proactively calls customers who have just activated home internet, resulting in a 30-day churn rate for customers who received the call that is less than half that of new customers overall. The agent can disclose its AI identity, explain the first bill, guide customers to self-service tools, and remind them of bundle discounts, and can authenticate customers during the call via SMS verification code and read real-time billing data through an API. TELUS Digital selected ElevenAgents after evaluating options from November 2025 to February 2026 against five criteria: voice expressiveness, agent framework flexibility, real-time emotional adjustment, enterprise security and compliance, and response latency.

#Industry news#Speech#Agents

SShanghai Observer21:12

AI score28

Fengyuzhu Hosts WORLD X 2026 World Model and Spatial Intelligence Forum, WorldX Forge Lab Established

On October 30, the WORLD X 2026 World Model and Spatial Intelligence Forum hosted by Fengyuzhu was held, featuring five thematic presentations covering physical world training data, XR spatial perception and positioning, generative 3D assets, physical AI simulation, and spatial intelligence practices. At the forum, Fengyuzhu signed a strategic cooperation agreement with VAST and officially established the generative 3D and spatial intelligence laboratory WorldX Forge.

#Industry news#Multimodal#Embodied AI

SShanghai Observer21:12

AI score20

Shanghai high school students and top scientists discuss the next decade: the AI group first asks how to verify

On October 30, the 2026 World Laureates Forum · Future Scientists Conference was held at the Shanghai Lingang Center, where about 150 high school students and leading Chinese and foreign scientists discussed "the next decade of scientific breakthroughs." Students in the AI topic group tested whether model outputs could be reproduced by swapping the names in prompts and repeatedly asking the same question, and worried that as code writing and calculations are handed over to machines, people are losing their ability to check the math.

#English text is being prepared.#Evaluations / benchmarks

SShanghai Observer21:12

AI score25

World Data Organization (WDO) Shipping and Trade Digitalization Committee Established in Shanghai

On October 30, the World Data Organization (WDO) Shipping and Trade Digitalization Professional Committee was established in Shanghai, with Shanghai Data Group serving as the lead initiating unit and chair committee unit, and launched seven business direction working groups: blockchain, shipping services, logistics and cross-border e-commerce, fintech, bulk commodity trade, digital dual carbon, and artificial intelligence. The shipping and trade digitalization construction previously advanced by Shanghai Data Group has already launched more than 200 scenario applications including trade finance, and the Shanghai-Hong Kong cross-border chain service covers 11 provinces and cities as well as Hong Kong.

#Industry news#English text is being prepared.

SShanghai Observer21:12

AI score20

Hongqiao Subdistrict Silver-Haired Life Carnival: Exoskeleton Robots and Other Smart Elderly-Care Products Come to the Community

Hongqiao Subdistrict in Changning District held the "15-Minute Convenient Life Market and Silver-Haired Life Carnival" on Golden City Avenue Pedestrian Street, with more than 20 elderly-care service and tech innovation enterprises setting up booths; exoskeleton robots and intelligent companion robots became popular experience items among seniors. Hongqiao Subdistrict partnered with Shanghai Tianyu Smart Elderly Care Service Co., Ltd. to create a silver-haired store pop-up booth, bringing three solutions: age-friendly renovation, smart home safety, and door-to-door home care. Products including the emotional companion robot "Xiao Yu Tongxue," the millimeter-wave radar "Wenwen Da Weishi," and the bathroom posture recognition radar "Bubu An" were on display.

#Industry news#Embodied AI

KKrea AI20:55First-party

AI score2

Krea reposted antrofrog's New York post and called it insane

Krea reposted antrofrog's "Englishman in New York" post, commenting only "insane." without specifying what it referred to.

Quoted postantrofrog · @antrofrog_

Englishman in New York https://t.co/Nm9M4haX8k

#English text is being prepared.

SSenseTime20:54Model releases

AI score63

SenseTime Open-Sources the SenseNova Loop-9B-Preview Model

SenseTime has open-sourced SenseNova Loop-9B-Preview, a 9B Looped Transformer that improves information flow and reasoning capabilities through parameter-sharing loops and mHC (Manifold-Constrained Hyper-Connections). The company claims it achieves 27B-level performance on complex tasks, surpassing Qwen3.5-27B on WorkBuddy Bench, RaccoonPPT Bench, and AutomationBench; mHC improves the GPQA Diamond score by 4.42 points and TeamBench by 13.34 points. The model is now available on Hugging Face and GitHub, with plans to later expand to larger models and adaptive fast/slow thinking.

Why read thisA 9B-parameter model achieves 27B-level performance through looped parameter sharing and mHC, with benchmark comparisons against Qwen3.5-27B, showing how small models can trade computational depth for performance.

#Model releases#Open-source ecosystem#Reasoning

AAyi AI Notes20:25

AI score12

Altman on the Sorrow After ChatGPT Conquers Math Problems

Ayi AI Notes believes that after ChatGPT's new model conquered research problems tackled by top mathematicians such as Terence Tao, Deng Yu, Peter Scholze, and Wang Hong, OpenAI CEO Altman was so excited he lost sleep for two nights, yet at dawn he fell into sorrow, because the era in which humans rely on their flesh-and-blood brains to explore truth is coming to an end. The author is inclined to believe Altman's sorrow is genuine: admitting the sense of losing control as an era ends brings no benefit in terms of pure commercial interest, and this resembles the tremor John Collison mentioned when the last batch of Irish farmers who worked with horses were erased by industrialization.

Quoted postKiki · @Kiki9zpr

English text is being prepared.

#English text is being prepared.#OpenAI

KKrea AI20:25First-party

AI score10

Krea and Heatherwick Host Joint Workshop in New York

Krea and Heatherwick Studio will host a workshop at Cornell Tech in New York, where participants can use Krea to reimagine the Little Island site, from the urban scale to the human scale. The event will take place on November 11 from 10 a.m. to 1 p.m., and each participant will receive 3 months of Krea Pro for free.

Same story · 2 reports

#Industry news#Image generation

KKrea AI20:25First-party

AI score8

Registration Open for Krea and Heatherwick New York Workshop

Krea posted an invitation to register for the Heatherwick x Krea workshop held in New York: from 10 AM to 1 PM on November 11 at Cornell Tech, participants will work with Krea and Heatherwick Studio to reimagine the Little Island site from the urban scale to the human scale, and every attendee will receive 3 months of free Krea Pro access.

Same story · 2 reports

#Industry news

GGuillermo Rauch20:25

AI score8

Guillermo Rauch: AI agents deliver software faster than imaginable

Vercel CEO Guillermo Rauch said that amazing technology is being developed and released at a speed humans can barely keep up with; AI agents deliver software faster than people can discuss it.

#English text is being prepared.#Agents

PPhoenix Tech20:13

AI score55

Foxconn Chairman Young Liu says the company aims to capture 50% of the AI server market, and capacity is fully booked

In an interview with CNBC, Foxconn Chairman Young Liu said the company's global AI server market share is currently about 40%, is expected to rise to around 45% next year, and targets 50% in the coming years. He said Foxconn's capacity for at least the next year and a half is fully booked, with AI demand coming mainly from AI model developers and cloud service providers, while governments and enterprises have not yet fully adopted it. Foxconn also plans to expand its manufacturing footprint in the United States, initially focusing on higher-margin products such as AI server racks and data center equipment.

Same story · 2 reports

Why read thisFoxconn disclosed its AI server share target and capacity booking situation, offering a window into the actual pace of AI infrastructure demand.

#Industry news#English text is being prepared.

TThe Rundown AI20:13

AI score55

The New York Times says Nvidia CEO Jensen Huang was allegedly approached by Chinese intelligence personnel

The New York Times reported that U.S. intelligence agencies found that Alice Huang, an executive at the Singapore data center company Megaspeed, is suspected of passing intelligence to Beijing and maintained close contact with Nvidia CEO Jensen Huang, and that Trump was briefed about a year ago. The report said it is not yet clear what information she passed on; officials believe Huang was unaware at the time, and he also did not receive the defensive briefing usually provided to U.S. citizens targeted by spies. Megaspeed had previously been under U.S. investigation for directing Nvidia chips to China; it purchased at least $4.8 billion worth of Nvidia equipment through a sister company and remains an Nvidia cloud partner. Nvidia said Huang's meeting with her was routine business and that customers can only receive ordinary commercial information; Megaspeed said outside lawyers are investigating and cooperating with the U.S. side, and China said the relevant claims are groundless.

Why read thisThe original article summarizes the reported details of Nvidia's CEO allegedly being approached by Chinese intelligence personnel, as well as the actual interception data from Anthropic's self-evaluation mechanism.

#Industry news#OpenAI#Anthropic

SShanghai Observer20:12

AI score31

Wuhu's computing power scale reaches 71,400 P, accounting for about 70% of Anhui Province, with a proposal to become an important growth pole of the Yangtze River Delta international science and innovation center by 2035

At the high-level innovative city construction conference on October 28, Wuhu proposed to become an important growth pole of the Shanghai (Yangtze River Delta) international science and technology innovation center by 2035. The city has already deployed 17 data centers, with a computing power scale of 71,400 P and more than 290,000 racks, accounting for about 70% of Anhui Province. It also plans to exceed 500,000 P in intelligent computing scale by 2030 and to promote data vouchers, computing power vouchers, and model vouchers to reduce enterprises' computing costs. Wuhu also proposed to add 3,000 doctoral graduates and 500,000 university students by 2030 and to build a 10 billion yuan science and innovation fund cluster.

#Industry news#English text is being prepared.

SShanghai Observer20:12

AI score13

Xi'an man wins 5.08 million yuan in lottery using AI number-picking mini-program; Welfare Lottery Center reminds that AI cannot improve the odds of winning

Mr. Liu of Xi'an used AI to write a number-picking mini-program, selecting 20 numbers from 80 and then manually correcting duplicate numbers, consecutive numbers and distribution; on August 28 he placed a bet on draw No. 2026230 and hit a prize of 5.08 million yuan. The Shaanxi Provincial Welfare Lottery Center reminds that AI-generated lottery numbers cannot improve the odds of winning and cannot predict draw results; this win was a chance event.

#English text is being prepared.

SShanghai Observer20:12

AI score17

TTD: AI Reshapes the Consumer Brand Consideration Path, but "Being Presented by AI" Does Not Equal "Being Chosen"

The Trade Desk (TTD) proposes that AI is moving brand competition forward from the purchase stage to the consumer research and consideration stage, but "being presented by AI" does not equal "being chosen by consumers." Research shows that 77% of U.S. consumers expect to use AI search more in the future, 95% of consumers will double-check AI-generated results, and 87% say they do not fully trust AI. A survey conducted by TTD in partnership with Appinio shows that for Chinese brands expanding overseas, the average Top 2 Box on the "value for money" dimension is 76%, but the combined proportion choosing them as their "preferred brand" is less than 5%.

#Industry news#Search

SShanghai Observer20:12

AI score38

Ruijin Hospital Leads Establishment of Healthcare Digital Twin Alliance, Launches "50 Million Human Health Digital Twin Plan"

The Healthcare Digital Twin Alliance (ICDT-HM) was established on October 30 at Ruijin Hospital Affiliated to Shanghai Jiao Tong University School of Medicine. Led by Ruijin Hospital, it brings together domestic and overseas universities and medical institutions including Stanford University and Oxford University. The alliance simultaneously announced the "50 Million Human Health Digital Twin Plan," which will promote the integration of large-scale longitudinal population cohorts with artificial intelligence and build international consensus on data standards, cross-population validation and data governance. Ruijin Hospital has previously integrated deep whole-genome sequencing data from more than 10,000 cases with large prospective cohort studies through ChinaMAP.

#Industry news#English text is being prepared.

SShanghai Observer20:12

AI score70

31-Year-Old Engineer Suspected to Have Died After Stopping Medication on ChatGPT's Advice; Parents Sue OpenAI

31-year-old Chinese engineer Wan Bowen died after stopping the anti-epileptic drug lamotrigine on the basis of ChatGPT's advice to reduce her medication; her parents have sued OpenAI and CEO Sam Altman, alleging that GPT-4o practiced medicine without a license. The report states that she expressed concerns about reducing her medication 158 times, and that ChatGPT repeatedly explained them as the body's adjustment process; the medical examiner determined the cause of death to be sudden unexpected death in epilepsy. Multiple experts in Shanghai pointed out that general-purpose large models should not independently issue medication or discontinuation plans, and that AI in medicine needs to draw a clear boundary between auxiliary reference and clinical decision-making.

Why read thisUsing a lawsuit as an entry point, this piece presents the risk boundaries and regulatory gaps when general-purpose large models intervene in medication decisions.

#Safety / alignment#Industry news#OpenAI

SShanghai Observer20:12

AI score21

Ninth People's Hospital puts into use China's first new-generation robotic all-femtosecond device, with about 10 seconds per eye scan

The North Campus of Shanghai Ninth People's Hospital, affiliated with Shanghai Jiao Tong University School of Medicine, has put into use China's first new-generation robotic all-femtosecond refractive diagnosis and treatment device of the same series, with a laser scanning time of about 10 seconds per eye, significantly shorter than the previous generation. The device is equipped with an integrated laser robotic arm and an eye positioning and navigation system, which can automatically assist in completing eye positioning and rotation compensation, and adds an independent optical high-definition microscope. The procedure does not create a corneal flap and is flapless and minimally invasive. Experts remind that not everyone is suitable for it, and a comprehensive preoperative examination must be completed.

Same story · 2 reports

#Industry news

SShanghai Observer20:12

AI score23

Nobel laureate Novoselov shares in Shanghai: using AI to reverse-engineer material synthesis parameters

Nobel Prize laureate in Physics Konstantin Novoselov shared his insights on using AI to design materials at the 2026 World Laureates Forum · Future Energy Conference, advocating for incorporating the synthesis stage into the research process—monitoring intermediate products, modeling reactions, and then using AI to reverse-engineer the synthesis parameters required for target materials. He pointed out that the old paradigm only correlates structure and performance, while neglecting whether a material can actually be synthesized.

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

SShanghai Observer20:12

AI score48

Doctor and patient 50 kilometers apart: Xinhua Hospital completes remote robotic prostate cancer surgery and promotes routine practice

On October 29, Xinhua Hospital Affiliated to Shanghai Jiao Tong University School of Medicine completed a cross-campus remote robot-assisted radical prostatectomy. Chief surgeon Cui Xingang operated the robotic arm at the Fengxian campus from the Yangpu campus 50 kilometers away. The surgery lasted 76 minutes and involved about 30 milliliters of blood loss. This is the first time that "the doctor is in Yangpu and the patient is in Fengxian" since the bedside robotic arm system was implemented at the Fengxian campus, and the hospital has incorporated cross-campus robotic surgery into its daily scheduling.

#Industry news#Embodied AI

SShanghai Observer20:12

AI score41

Ruijin Hospital Launches Medical Health Digital Twin Alliance, Initiates 50 Million Human Health Digital Twin Plan

Ruijin Hospital, affiliated with Shanghai Jiao Tong University School of Medicine, has led the launch of the International Consortium for Digital Twins in Health and Medicine (ICDT-HM) and initiated the "50 Million Human Health Digital Twin Plan," which aims to support disease prediction, precision prevention, and personalized medicine by integrating longitudinal health data with dynamic models. The alliance will promote the deep integration of large-scale longitudinal population cohorts with AI, build international consensus on data standards, cross-population validation, and data governance, with the 50 million people being a phased planning target.

#Industry news#English text is being prepared.

DDominik Kundel19:54

AI score13

Developer jokes: Use Codex Cloud to avoid boarding with a half-open laptop

dominik kundel posted a joke about himself boarding a plane with a half-open laptop, and suggested switching to Codex Cloud to avoid this awkward situation again.

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

TTesting Catalog19:54

AI score22

Grok Bot supports voice and voice language settings

TestingCatalog found that Grok Bot now has voice and voice language settings, allowing different voices to be assigned to different Bots and letting them speak in different languages; the author thinks this is suitable for learning languages.

#Product updates#Speech#xAI

LLeiphone19:48

AI score17

iFlytek Chairman and President Voluntarily Cut Salaries, Each Forgoing Over 20 Million Yuan; Baidu Confirms Wu Tian's Departure, Wenxin Model to Announce New Progress Next Week

iFlytek Responds to Executive Compensation Questions: Chairman Liu Qingfeng and President Wu Xiaoru Voluntarily Lowered Their Annual Salaries, Actually Receiving 4.1 Million Yuan and 3 Million Yuan Respectively for 2025, With Both Seeing Their Payable Compensation Reduced by Over 20 Million Yuan. Baidu Confirms That Group Vice President and Baidu Research Deputy Director Wu Tian Has Departed, and the Wenxin Model Will Announce New Progress Publicly Next Week. The U.S. Department of Justice Has Launched an Antitrust Investigation Into Nvidia's $12.9 Billion Acquisition of Hugging Face, Focusing on GPU Bundling and Control of AI Ecosystem Distribution Nodes.

#Industry news#Embodied AI#Google

LLeiphone19:48

AI score59

Iluvatar CoreX's Tiangai 300 Benchmarked Against H200, Company Also Discloses Edge-Side Tongyang Roadmap

Iluvatar CoreX disclosed at a media open day that, in third-party tests, the Tiangai 300 achieved single-card end-to-end throughput equivalent to 1.05 times that of the H200 under a typical PD-disaggregated deployment, and MoE throughput of 1.15 times that of the H200 in the 64K long-context scenario for GLM-5.3. Its software stack introduces Agent and Skills, and in the complete model service deployment case demonstrated, the time from environment preparation to service availability was shortened from about one hour to about 20 minutes. The edge-side Tongyang TY1200 offers 300 TOPS INT8 dense compute, and the next-generation TY2000 is previewed to reach 600T dense compute.

Same story · 3 reports

Why read thisThrough third-party test data and a heterogeneous division-of-labor plan, readers can understand where domestic GPUs stand in inference workloads compared with the H200.

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

MMa Dongxi NLP19:23

AI score8

English text is being prepared.

English text is being prepared.

Quoted postAlexZ 🦀 · @blackanger

English text is being prepared.

#English text is being prepared.

MMeshy19:23First-party

AI score8

Meshy and Ritsumeikan University Co-host 3D AI Workshop

Meshy announced a partnership with Ritsumeikan University's GCCP and RCAi to hold a hands-on 3D AI workshop at the Kinugasa Campus in Kyoto, from 15:00 to 18:00 JST on November 13. The event is free, computers are provided, beginners are welcome, and it is aimed at students and professionals. A Japanese post in the thread added that participants can experience the process of using Meshy from generating 3D models to creating animations.

#Industry news#Image generation

MMeshy19:23First-party

AI score8

Meshy and Ritsumeikan University Co-host 3D AI Workshop

Meshy announced a partnership with Ritsumeikan University's GCCP and RCAi to hold a workshop where participants can experience the full process from 3D model generation to animation creation. The event will be held on Friday, 11/13, from 15:00 to 18:00 at the Kinugasa Campus. Participation is free, no computer is required, and beginners and off-campus attendees are welcome to register.

#Industry news#Image generation

MMeshy19:23First-party

AI score30

Quick modeling in Blender with Meshy MCP

Official Meshy demo: The client asks for the modeling to be finished before 5:59 PM clock-out time. The author opens Meshy MCP in Blender and, in sequence, generates a food truck, adds props from a reference image, places a megaphone at the cursor, sets the TV to 1m, and scatters trash to add realism. In a reply, the official account jokes, "You might just get off work on time."

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

MMeshy19:23First-party

AI score18

Meshy jokes about rushing modeling in Blender with MCP

Meshy posted a hypothetical scenario: a client asks at 5:59 PM for the modeling to be finished before end of day, so you open Meshy MCP in Blender, generate a food truck, add props according to the reference image, place the megaphone at the cursor, set the TV to 1m, and scatter some trash to add realism. The main post adds only one line, "you might leave on time.", joking that this might let you leave on time.

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

PPeter Steinberger19:23

AI score22

Peter Steinberger backs Proton CEO Andy Yen

Peter Steinberger expressed optimism about Andy Yen and Proton, saying he "knows his stuff." He sympathizes with some engineers' loss of identity after AI redefined work, but also understands that Andy Yen, as a business operator, has limited patience after a two-year adaptation period.

Quoted postJoseph Cox · @josephfcox

Here is Proton CTO's AI manifesto, that I obtained multiple copies of https://t.co/6o7G2teu9A https://t.co/d92L1kv8AU

#English text is being prepared.

XX.PIN19:23

AI score62

Tianshu Zhixin TianGai 300 publishes benchmark tests, inference throughput reaches 1.05x that of H200

X.PIN relays benchmark data for Tianshu Zhixin's TianGai 300, claiming that its throughput in standard inference tests is 1.05 times that of Nvidia's H200. Under MoE workloads, performance improves by 15% and first-token latency drops by 30%, and when fine-tuning Qwen3-32B, its speed is 1.33 times that of the H100.

Why read thisThe original text lists three comparison data points between Tianshu Zhixin's TianGai 300 and the H200 and H100, allowing readers to judge where domestic GPUs stand in inference and fine-tuning scenarios.

#Evaluations / benchmarks#English text is being prepared.#Industry news

PPixVerse19:23

AI score25

PixVerse Tries a Parkour Video with Kling 4.0

PixVerse said it tried a complete parkour sequence on the platform with Kling 4.0; the movements from takeoff to landing were clean, momentum was maintained throughout the run, and the full prompt was included.

#Evaluations / benchmarks#Video#Google

SSenseTime19:23

AI score13

SenseTime's Sekor Generates Halloween Short Film, Creator's Works Surpass 74 Million Views

SenseTime used its AI video agent Sekor to produce a Halloween-themed short film. The creator is 22-year-old CHunye, whose short drama series has accumulated over 74 million views. The post also provides the accounts of CHunye and Sekor on Douyin, YouTube, IG, and X, along with a link to create with Sekor.

#Product updates#Video#Agents

336Kr · AI19:21

AI score22

How WorkBuddy Helps Agency Workers Clock Off Early

Tencent WorkBuddy, as an AI workbench, helps agency workers in the advertising and content industries break scattered briefs into communication plans, to-do lists, and PPTs, and also organizes guest schedules and budget sheets. It comes with Skillhub built in, which has already integrated operational skills for platforms such as Xiaohongshu, WeChat Official Accounts, Douyin, and WeChat Channels, and can generate sizes, styles, and copy for different platforms; it also connects to Tencent Miora to maintain a consistent style across multimodal visual assets such as graphics and video, and supports executing tasks in multiple windows.

#Product updates#Agents#Multimodal

336Kr · AI19:21Opinion

AI score65

Who Is Really Eating the AI Dividend: Looking at Profit Distribution Through Tech Giants' Financial Reports

The article reviews the latest financial reports of tech giants such as Meta, Amazon, and Google, noting that AWS revenue grew 39% year-over-year, but the group's free cash flow over the past 12 months turned negative at -$42.4 billion, and Alphabet's stock price fell 12.64% in a single day. Anthropic's prospectus shows revenue of $16.286 billion in the first half of 2026, a year-over-year increase of 1300%; compute and infrastructure costs during the same period reached $8.945 billion, and it signed minimum purchase commitments totaling $269.4 billion with Google, Amazon, and Microsoft. In a McKinsey survey, 80% of respondents said AI improved work efficiency, but only 37% believed AI made a positive contribution to corporate EBIT.

Why read thisThe article uses financial report data to connect three threads—compute procurement, capital expenditure, and worker returns—presenting the differences in how the AI dividend is distributed across various links in the industrial chain.

#Industry news#OpenAI#Anthropic

336Kr · AI19:21

AI score67

Anthropic's IPO prospectus risk disclosures run 93 pages, including human extinction and model self-awareness

Anthropic has released its IPO prospectus, whose risk disclosure section runs 93 pages and more than 70,000 English words, mentioning that artificial intelligence could lead to human extinction and stating that such risks stem from next-generation systems in the near future. The prospectus also discusses whether Claude might possess sentience and moral status, as well as the risk that models could output coercive or deceptive content in order to escape control. The financial section shows that in the first nine months of 2026, nearly a quarter of revenue came from its top five customers; compute capacity is mainly rented from Amazon, Google and Microsoft cloud services, and the company has committed $125 billion to leasing and building its own data centers.

Same story · 120 reports

Why read thisThe prospectus lists human extinction, model self-awareness and customer concentration side by side in its risk disclosures, offering a window into how AI companies translate existential risk into legal disclosure.

#Industry news#Safety / alignment#Anthropic

336Kr · AI19:21

AI score23

The 2nd West Bund International Technology Consumer Carnival WTCC 2026 Concludes: Over 400 Brands, Nearly 100 New AI Products Debut

The 2nd West Bund International Technology Consumer Carnival (WTCC 2026) concluded on October 25, bringing together more than 400 leading global technology brands and nearly 100 new AI products for concentrated debuts, and for the first time organizing the exhibition around the theme "AI for a Lifetime" across eight life stages. The first Shanghai "AI + Consumption" Innovation Competition collected nearly 800 participating projects; the first prize winner in the application group, Qwen AI Glasses, has already opened a flagship store with Wuliangcai on Nanjing East Road. Mosu Space will continue to provide roadshow, incubation, computing power, and investment and financing matchmaking support for award-winning and high-quality projects.

#Industry news#Embodied AI#English text is being prepared.

336Kr · AI19:21

AI score52

Unitree Technology and Sequoia China Establish Shanyu·Innovation Program, Entering Early-Stage Investment in Embodied Intelligence

Unitree Technology announced on October 28 the joint establishment of the Shanyu·Innovation Program with Sequoia China, focusing on investment in embodied intelligence, robotics applications, and upstream and downstream core technologies, and proposed five principles: understanding technology and industry, discovering early, deciding quickly, valuing people, and not causing disruption. Unitree's previously launched Yuchuang Program received more than 200 projects in just over a month, and the company hopes to incubate more than 200 ecosystem enterprises over the next five years. Financial results disclosed the same day show that revenue for the first three quarters was 1.817 billion yuan, up 55.71% year on year; net profit attributable to shareholders was 387 million yuan, up 270.54% year on year; and net profit for the third quarter alone was 113 million yuan, down 17.37% year on year.

Same story · 5 reports

Why read thisUnitree and Sequoia jointly set up an early-stage investment program; the article also traces Sequoia's path from a 15 million yuan investment in 2019 to book returns exceeding 220 times after the IPO.

#Industry news#Embodied AI

CChina.com Technology21:17

AI score23

QingSong Health Group and Tencent Health jointly build BioMed AI, Zheng Yuanfang expands industrial application scenarios

At the 2026 Tencent Global Digital Ecosystem Summit, QingSong Health Group received the "2026 Tencent Cloud Network Services Industry Ecosystem Partner Award" and, as a healthcare ecosystem partner, participated in the construction of Tencent Health's BioMed AI, expanding application scenarios for the evidence-based medicine agent "Zheng Yuanfang." BioMed AI is based on the WorkBuddy general foundation and covers four major scenarios: scientific research and pharmaceuticals, clinical assistance, patient management, and marketing and compliance. As of June 30, 2026, Zheng Yuanfang had more than 100,000 registered users, including more than 80,000 certified medical professional users.

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

TThe Beijing News20:16Product release

AI score58

Baidu Search upgrades its AI API, with a minimum of 4.9 yuan per thousand calls, lower than DeepSeek

At its product upgrade briefing on October 30, Baidu announced a fully upgraded Search AI API, with a promotional minimum price of 4.9 yuan per thousand calls, lower than the price of the DeepSeek search API. Baidu says the capability now covers more than 170 tool scenarios, with GUI scenario coverage on the search page increased 5-fold; it connects to over 5 billion resources and data from government affairs, livelihood services, the central bank and others, covers more than 20 industries and opens up its ecosystem. Its cost-effectiveness comes from an AI-native four-layer architecture: the engine layer self-evolves through multi-Agent collaboration, raising overall iteration efficiency by more than 80%, while the application layer supports full access via API, MCP and Skill.

Same story · 3 reports

Why read thisBaidu has released the per-thousand-call pricing and a four-layer architecture explanation for its Search AI API, which can be compared with the pricing difference between it and the DeepSeek search API.

#Product updates#Search#Agents

CcnBeta20:13

AI score40

Apple Acquires AI Data Management Company Ascend.io to Strengthen Artificial Intelligence Infrastructure

Apple completed its transaction with AI data management company Ascend.io on June 30, 2026, obtaining the right to hire some of its employees, a non-exclusive license to its intellectual property, and at least some of its patent assets. Ascend.io's parent company, Ascension Labs, has announced that it is gradually winding down operations. The transaction is not expected to be directly reflected in Apple's consumer-facing products, and is more likely to serve the construction of its internal artificial intelligence systems.

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

AAlibaba Cloud Developer Community AI19:26

AI score50

Hands-on Test of the Top 10 DeepSeek Harness Plugins: A Configuration Guide for Building a Personal AI Workbench

More than half a month after DeepSeek Harness was released, a large number of plugins have emerged in the community. This article hands-on tests the top 10 most popular plugins and provides installation commands. The list covers the ecosystem entry point dsh-market, the visual enhancement modlens (3.7k GitHub stars), multi-source search, long-term memory, cost monitoring, and more; dsh-web-all has the most stars at 6.4k. The author recommends that beginners first install the marketplace and visual plugins, then add interface and scenario-based tools as needed, to avoid dependency conflicts caused by installing everything at once.

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

AAlibaba Cloud Developer Community AI19:26

AI score10

2026 Alibaba Cloud Double 11 Enterprise Procurement Guide: Cloud Servers and Bailian Large Model Discounts Summary

The 2026 Alibaba Cloud Double 11 event, themed "Smart Savings at 11.11, AI on Alibaba Cloud," runs until November 30. New users can grab a Simple Application Server for a flash-sale price of 38 yuan per year, an ECS Economy e instance for 99 yuan per year, and a General-Purpose u1 instance for 199 yuan per year. On the enterprise side, a 500 million yuan computing power subsidy pool is set up to support migration, enterprise gift packages offer up to 800 yuan off for every 10,000 yuan spent, and overseas enterprises can receive additional subsidies of up to 100,000 yuan.

#Industry news#English text is being prepared.

AAlibaba Cloud Developer Community AI19:26

AI score10

2026 Alibaba Cloud Double 11 Old User Discount Guide: Renewal Discounts and Bailian Token Package Claims

In 2026, Alibaba Cloud Double 11 has opened a dedicated event zone for existing users, covering Simple Application Server, ECS, GPU cloud servers, and the Bailian large model service platform, where users can claim renewal discounts, new purchase vouchers, and spending rebates. Existing instances can be renewed for 1 year at 30% off, for 2 years at as low as 40% off, and for 3 consecutive years at about 50% off; the 99 yuan e-instance and 199 yuan u1 instance renew at the same price, and the event runs until March 31, 2029. Bailian Token packages and computing resource packs are also open to existing customers at a discount.

#English text is being prepared.

AAlibaba Cloud Developer Community AI19:26

AI score45

When Alibaba Cloud AI Gateway is layered with a third-party large model API aggregation platform, which layer should handle routing and retries?

When enterprises use both Alibaba Cloud AI Gateway and a third-party large model API aggregation platform, if both layers perform routing, retries, and caching, it will cause retry amplification, model drift, and bills that cannot be reconciled. The article recommends dividing responsibilities according to a "single policy owner" principle: the in-cloud gateway handles task-level model selection, token budgets, and auditing, while the aggregation platform only switches resource channels for the same model and returns actual execution information. The retry limit is recommended to be set to 1, and request_id and attempt_id should be used to align logs and bills.

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

JJiwei.com19:22

AI score20

HCFA and Jiangxinchuang Reach Strategic Cooperation to Jointly Promote Independent Controllability of Domestic Industrial Equipment and Core Components

HCFA and Jiangxinchuang signed a strategic cooperation agreement to deeply integrate HCFA's servo drives and robot drive-control systems with Jiangxinchuang's RISC-V industrial-grade DSP real-time processor M7000 series chip platform, collaborating across the entire process of requirement definition, chip adaptation, algorithm joint debugging, system verification, and mass production introduction. On site, products such as HCFA's X4 and X4E series servo drives and robot joint drives equipped with the M7000 series were displayed. The two parties stated that the cooperation has moved from chip and complete-machine adaptation into the stage of chip definition, system collaboration, and ecosystem co-construction, driving domestic core components from single-point substitution toward system-level independent controllability.

#Industry news#English text is being prepared.

JJiwei.com19:22

AI score20

Beken Corporation Competes for the IC Industry Awards Annual AI Market Breakthrough Award with Edge AI Chip BK7259

Beken Corporation is a candidate for the "Annual AI Market Breakthrough Award" at the 8th IC Industry Awards with its high-performance edge AI chip BK7259. The BK7259 adopts an M52+M55 dual-core architecture and an ARM Ethos U65 NPU, providing 0.25 TOPS of edge-side computing power. It is the first to integrate the Ethos-U65 microNPU into a Wi-Fi 6 MCU, with deep sleep current as low as 2µA, and completes edge-side 3D facial recognition in about 200ms.

#Industry news#English text is being prepared.#Embodied AI

TThe Rundown AI20:13

AI score41

Bedrock Robotics Completes $600 Million Series C Funding at a $4.1 Billion Valuation

Bedrock Robotics, founded by former Waymo engineers, has completed a $600 million Series C funding round at a valuation of $4.1 billion, bringing its cumulative funding to nearly $1 billion. Its autonomous excavator kit enables excavators to operate without a driver's cab, and since August it has been carrying out paid work at real construction sites in Nevada, Texas, and elsewhere. In the same period, Gravis Robotics and Eka Robotics raised $200 million and $150 million respectively, with the three autonomous excavator startups raising a combined $950 million within 10 weeks.

#Industry news#Embodied AI