The latest AI news, in focus.

From Large Language Models to World Models

Yan Deli, a senior expert at Tencent: LLMs are probabilistic models. They do not truly understand language; rather, they predict the next token (Next Token Prediction). Essentially, they are monotonous “word-chain” machines—moving forward one step at a time without retracing their path. At their core, LLMs are probabilistic models that learn, from statistical co-occurrence patterns, “what often appears alongside what,” and output “this is usually the case,” without grasping “why it is so.”...

After reading the prospectus, I finally understand why Anthropic no longer cares about its users.

The big one has finally arrived. On October 26, Anthropic publicly filed its S‑1 registration statement with the U.S. Securities and Exchange Commission, preparing to list on the Nasdaq Global Select Market under the ticker symbol ANT. Media reports suggest the company’s valuation is being discussed in a range of $1.5 trillion to $2 trillion, though the proposed offering price range remains undisclosed. Within 24 hours, this more than 300-page prospectus had already been scrutinized inside and outside the tech and finance communities, with everyone focusing on the most eye‑catching figures: revenue of $16.286 billion in the first half of 2026, compared to just $1.163 billion in the same period last year; operating profit of $974 million in the second quarter, marking the company’s first-ever quarterly operating profit; and a net loss of $80.7 billion, largely driven by $72.388 billion in changes to the fair value of convertible bonds, derivative liabilities, and SAFE notes—non‑cash accounting expenses. Yet beyond these Wall Street narratives, domestic developers and everyday users experience something quite different. In the eyes of mainland Chinese users, Claude is arguably the AI product currently facing the most account bans, the strictest risk controls, and the most frustrating user experience. Why does a company that has built a state‑of‑the‑art large model consistently appear so indifferent when dealing with ordinary individual users? After poring over the financial footnotes and risk disclosures in the prospectus, the answer is actually right there in the numbers. In Anthropic’s business landscape, personal users contribute only a very small share of its revenue. 01 Less Than 20% from Individual Users To understand who a company truly prioritizes, the simplest approach is to examine its revenue breakdown. In Note 2 of the financial statements attached to the prospectus, Anthropic splits its revenue into two main categories: usage‑based revenue and subscription revenue. The contrast is stark: for the full year 2025, usage‑based revenue totaled $3.796 billion, while subscription revenue was just $789 million—roughly an 83% to 17% split. By the first half of 2026, usage‑based revenue surged to $13.064 billion, with subscription revenue at $3.222 billion, still leaving usage‑based revenue accounting for about 80% and subscriptions making up roughly 20%. Even more telling details lie hidden within the composition of subscription revenue. Of that 20%, there are both individual users’ Cl

b11923

ggml : relax scale to non-contiguous views (#31146) * ggml : relax scale to non-contiguous views SCALE requires src0 to be 1D-contiguous. Relax this to dim-0 contiguity with arbitrary row strides (ggml_is_contiguous_rows), same as the unary ops in the Metal backend. - ggml: relax the assert in ggml_scale_impl - cpu: address src0 rows through their byte strides (f32, bf16) - cuda: stride-aware kernel, fastdiv row decomposition - metal: support bf16 scale (guarded unary instantiations) - sycl: keep the gate at contiguous, the kernel is flat f32 Add view variants to test_scale. Verified on CPU, Metal and CUDA. Assisted-by: dsh:llama.cpp/ggml-org/Qwen3.8-Flash-Next-GGUF (RTX 5090 32GB, DDR5 128GB) * cuda : skip the softcap fusion on non-contiguous views The fused softcap kernel reads src and w

Manus has just quietly released its 2.0 manifesto.

On October 26, Zhang Tao (online alias HideCloud), co-founder and product partner at Manus, recently published a lengthy article on his WeChat public account “Qianyun Sisi,” titled “Transforming Manus with Manus.” The piece opens with an image that left a deep impression on him. He had submitted several requirements and bug reports to an engineer; when he checked back later, he found that the engineer’s computer had multiple Codex and Manus sessions open. Clicking on any one of them, he saw that the input was precisely the original requirement he’d just posted on Slack—without a single word changed. Even more surprising to him was that this engineer was simultaneously working on over ten or twenty tasks. At Manus today, this has become the norm. The article is packed with information, but a few key points are especially worth noting. Internally, Manus has created a unified code entry point called Manus-mono, accompanied by a manual designed for AI, enabling the AI system to view the entire project landscape across repositories. With a single request, one person can complete all client-side development. The mobile team has consistently consisted of only seven members, yet they manage two products and four apps, keeping their progress largely in sync with the web version. All product managers have submitted code to Manus, and roughly half of them can independently bring their requirements to production. Some product managers even turn their written requirements into video presentations. Meanwhile, the marketing team now has a “digital colleague” that schedules its own tasks, proactively reminds everyone of project deadlines, and even reaches out across teams to clarify feature details. In September, the company’s per-capita AI spending exceeded $10,000, with a small number of individuals consuming as much as $70,000 to $80,000 in a single month. Zhang Tao dedicated an entire section to addressing external skepticism. The timing of the article’s release is also noteworthy. At the end of September, Manus launched version 2.0 along with its personal AI agent app, Cue; in early October, the parent company Butterfly Effect announced the completion of a new round of financing exceeding $500 million, while simultaneously opening 17 positions in Beijing. Seen against this backdrop, the article reads less like a routine update and more like a manifesto marking the Manus team’s entry into the 2.0 era. It discusses how the organization leverages AI to evolve, while revealing the company’s product roadmap, cost structure, and hiring standards. 01 Manus itself serves as a showcase for 2.0 If you compare the internal practices described in the article with the feature list of Manus 2.0, you’ll find…

Anthropic still has to keep “charging ahead”

The prospectus shows that in the first half of 2026, the company generated revenue of US$16.286 billion, roughly 14 times that of the same period last year, while its operating loss narrowed from US$2.135 billion to US$693 million.

n8n@3.0.0

# [3.0.0](https://github.com/n8n-io/n8n/compare/n8n@2.45.0...n8n@3.0.0) (2026-10-27) ### Bug Fixes * Add file-type tooltip to artifact download button ([#41490](https://github.com/n8n-io/n8n/issues/41490)) ([4a74f02](https://github.com/n8n-io/n8n/commit/4a74f027e612f31098b354dd3a77b2c2ab98281c)) * **ai-builder:** Check execution:read in Assistant execution reads ([#41863](https://github.com/n8n-io/n8n/issues/41863)) ([a967486](https://github.com/n8n-io/n8n/commit/a9674866d075e989a5ce896432262c9a7dc8b0c7)) * **ai-builder:** Send Assistant model calls through the configured AI transport ([#41948](https://github.com/n8n-io/n8n/issues/41948)) ([9b63561](https://github.com/n8n-io/n8n/commit/9b63561cbb48f06d216f27be3b77722d97e4e74a)) * Allow custom OAuth2 scopes ([#41249](https://github.com/n8n-

Qiyuan Robotics’ 5,000th unit has rolled off the production line, marking a transition from large-scale mass production to commercial validation.

On October 25, Qiyuan Robotics officially rolled off the production line its 5,000th personal robot. From the launch of the Qiyuan Q1 and T1 personal robots on September 20 to reaching this milestone just one month later, Qiyuan has made significant progress in ramping up production capacity, optimizing its manufacturing system, and enhancing supply-chain coordination. For the personal robotics industry, which remains in its early stages of commercialization, the significance of 5,000 units extends beyond mere volume growth. It represents the company’s ability to translate R&D‑driven technological expertise into replicable manufacturing capabilities, while also seamlessly integrating product delivery, user experience, and software iteration—key steps that bridge the gap between technological exploration and consumer‑market adoption, and crucial metrics for assessing commercial viability. As artificial intelligence moves from the digital realm into the physical world, competition in the robotics sector is expanding beyond technology showcases to encompass engineering, productization, and commercial execution. Qiyuan’s recent mass‑production effort provides a concrete case study for observing the industrialization trajectory of personal robotics. 01: From Zero to 5,000 Units in One Month—A Robust Mass‑Production System Takes Shape “Since the production line went live, it took only one month to move from zero to consistently producing and shipping 5,000 units,” said Gao Quanzhong, Co‑President of Qiyuan Robotics and President of the Central R&D Department. Currently, the line produces one robot every 2.5 minutes, with daily output increasing from just over 10 units to more than 200. For personal robots, achieving mass production involves far more than simply adding lines or speeding up assembly. These devices integrate multiple systems—including mechanical architecture, motion control, sensory interaction, and intelligent algorithms—requiring both sophisticated functionality and precise assembly tolerances, consistent quality, and reliable performance at scale. Evaluating true mass‑production capability demands more than just counting finished units; it requires assessing whether the manufacturing system can sustain continuous output, whether quality standards can support large‑scale replication, and whether the supply chain can meet delivery demands. According to the company, to address these challenges, Qiyuan has implemented process standardization, quality‑control improvements, automated assembly, and calibrated testing to enhance batch consistency. On the supply‑chain side, it has collaborated with suppliers to establish centralized raw‑material warehouses, adopted semi‑automated assembly lines, and built an end‑to‑end manufacturing IT system that ensures full traceability throughout the production process. Additionally, Qiyuan has established rigorous validation protocols—including million‑cycle wire‑harness bend tests, hundred‑thousand‑cycle joint durability tests, and module impact‑resistance assessments—to bolster long‑term stability under real‑world usage conditions. Collectively, these capabilities form the engineering foundation for transitioning from prototype development to scalable manufacturing. The rollout of 5,000 units still…

An agent that helps users snatch up iPhone foldable-screen models is reshaping e-commerce in the AI era.

Shopify AI saw transaction orders grow by a factor of 13. In the early morning of October 16, pre-orders began for Apple’s first foldable-screen phone, the iPhone Duo. Priced from $1,999, initial production capacity was tight, and delivery dates quickly slipped to January next year. This time, it wasn’t just people glued to their screens vying for the launch. This year, Agents have become the strictest “parents” to iPhone Duo scalpers. | Image source: Apple On the MacRumors forum, one user who missed out vented their frustration, while on Reddit others boasted about using Muse bots to snatch up multiple units in a single go—before even a human could click, the inventory had already been swept away by AI. Muse is Meta’s personal Agent, launched on September 8; shortly after its debut, it topped the free charts in the U.S. App Store, with downloads soon surpassing 2.5 million. According to Meta, shopping has become one of Muse’s primary use cases. Around the time the iPhone Duo went on sale, Muse even released a new version optimized for this device, with an update note that emphasized making in‑chat shopping more convenient. AI-driven shopping has moved from demo stages at product launches straight into real‑world stock‑picking scenarios. But not every retailer welcomes these new customers. In late September, Amazon began banning Muse. By October, users of Musk’s Grok Bot encountered the same warning when placing orders on Amazon: unauthorized AI agents continuing to access the platform violates Amazon’s terms of service. Meanwhile, another company, Shopify, took the exact opposite approach. On September 21, Shopify CEO Tobi Lütke announced a deep partnership with Muse, enabling Muse to process checkout through Shop Pay across all Shopify stores. And Shopify’s financial reports already show early returns from this strategy: in the first quarter of this year, orders generated via AI search grew nearly thirteenfold year over year, while store traffic driven by AI increased eightfold. Closing doors on one hand, opening them on the other—when more and more purchasing decisions are handed over to Agents, will the question of “whether or not to let Agents in” become…

n8n@2.46.0

# [2.46.0](https://github.com/n8n-io/n8n/compare/n8n@2.45.0...n8n@2.46.0) (2026-10-27) ### Bug Fixes * Add file-type tooltip to artifact download button ([#41490](https://github.com/n8n-io/n8n/issues/41490)) ([4a74f02](https://github.com/n8n-io/n8n/commit/4a74f027e612f31098b354dd3a77b2c2ab98281c)) * **ai-builder:** Check execution:read in Assistant execution reads ([#41863](https://github.com/n8n-io/n8n/issues/41863)) ([a967486](https://github.com/n8n-io/n8n/commit/a9674866d075e989a5ce896432262c9a7dc8b0c7)) * **ai-builder:** Send Assistant model calls through the configured AI transport ([#41948](https://github.com/n8n-io/n8n/issues/41948)) ([9b63561](https://github.com/n8n-io/n8n/commit/9b63561cbb48f06d216f27be3b77722d97e4e74a)) * Allow custom OAuth2 scopes ([#41249](https://github.com/n8

OpenAI has rewritten the history of human science! “The King’s Conjecture” has been disproven on all fronts, and the 30-year-old ironclad rule governing ETH has collapsed.

New智元 reports: Honestly, the history of human science has been completely rewritten tonight! Just now, OpenAI dropped three major announcements in quick succession, officially revealing that its internal AI independently achieved three groundbreaking discoveries—each a true “milestone”: In mathematics, it proved the “local smoothness conjecture” in all dimensions, finally putting an end to the “Kakeya conjecture,” which had puzzled the mathematical community for a century; in computer science, it refuted the “Exponential Time Hypothesis” (ETH), which has dominated the field for nearly 30 years; and in theoretical physics, it provided insights into four-dimensional quantum chromodynamics (QCD) and pure...

Neocloud: This wall on the verge of collapse—Jensen Huang is propping it up, while Musk is pushing it down.

Jensen Huang is still struggling to prop up the Neocloud wall, while Musk has already predicted its collapse. The first domino in the AI bubble may well be among these companies. On October 25, Musk issued a warning on X: in recent weeks, the debt markets have virtually stopped extending financing to firms that lack substantial revenue streams beyond their AI data centers. He forecast that more than 200 Neocloud players will face “guppy‑level” deaths and consolidation, with only a handful ultimately surviving. Yet just three days earlier, NVIDIA CEO Jensen Huang offered a nearly opposite assessment. On October 22, during an interview with CNBC, Huang insisted that demand for Neocloud computing power remains extremely robust; the real bottleneck is insufficient working capital, and funding challenges will eventually be resolved. Both men remain optimistic about the demand for AI compute, but they differ sharply on how to manage the business of renting out computing power...

In the era of large models, who will become that “iPhone” no one is willing to replace?

The champion on the AI rankings may change, but the products users rely on every day don’t necessarily. The next round of competition in large models will hinge not just on whose model is stronger, but on turning a single use into a habit.

b11922

tests : fix multi-seq split replay and partial-state cases for trained models (#31216) * tests : fix multi-seq split replay against trained models The rollback snapshot is captured by the graph that decoded the rolled-back tail, while the reference context decodes the same prefix with different batch shapes. The two states differ in the last bits and trained models amplify the difference with every replayed token. Compare only the first replayed token of each seq, then restore the rollback state into the reference context before the seq-1-only tail. Assisted-by: dsh:llama.cpp/ggml-org/Qwen3.8-Flash-Next-GGUF (MacBook, M5 Max, 128GB) * tests : require bit-exact seq-1 isolation in split replay The seq state save is not always bit-lossless - SWA-masked cells are not saved - so reload the rest

MiniMax M3.1Flash release sparks “questionable results” concerns: high scores on old questions, failure on new ones, and an illusion rate as high as 92.8%

On October 19, 2026, MiniMax released its low-cost large model, M3.1Flash, which scored 42 points in comprehensive evaluations and boasts a 70% reduction in cost compared to the previous generation. However, it immediately faced accusations of “cheating on exam questions”: on AA’s publicly available past papers, it nearly achieved a perfect score, yet continued to produce the original answers even after the question stems were rewritten. Reviewers such as Cat Boss and Binkg pointed out that its performance on old and new questions is sharply divided, token consumption is abnormal, and its hallucination rate in knowledge tests reaches as high as 92.8%. On October 23, MiniMax open-sourced the model weights for independent verification, but the mechanisms behind its question-identification capabilities and the reasons for discrepancies in Chinese–English responses remain unresolved.

Why it mattersThis report is based on multi‑party empirical testing (including Binkg’s re‑title experiment, the author’s replication, and a dual evaluation comparison between AA and Plotly‑Mila), verifiable data (token counts, prices, accuracy/hallucination rates), and contradictions in model behavior (English errors vs. Chinese correctness, stable performance on old tasks vs. collapse on new ones). It systematically lays out the core symptoms of the current credibility crisis in large‑model evaluations: high scores do not necessarily equate to strong generalization, and low‑price advantages create tension with the reality of model capabilities. The information source is a deep investigation authorized for publication by Titanium Media, complete with original test procedures, screenshot evidence, and timestamps, offering both timeliness and rich detail.

The era of profiting from token flipping is over—what few viable paths remain for AI intermediaries?

AI Hub seeks new avenues.

b11921

llama : remove obsolete n_pad from memory instances (#31200) Every caller passes n_pad = 1 and the only two users clamp it to 256. Drop the parameter, the members and the trivially true kv_size assert. Assisted-by: dsh:llama.cpp/ggml-org/Qwen3.8-Flash-Next-GGUF (MacBook, M5 Max, 128GB) **Website:** - **Attestations:** - **macOS/iOS:** - [macOS Apple Silicon (arm64)](https://github.com/ggml-org/llama.cpp/releases/download/b11921/llama-b11921-bin-macos-arm64.tar.gz) - macOS Apple Silicon (arm64, KleidiAI enabled) [DISABLED](https://github.com/ggml-org/llama.cpp/pull/23780) - [macOS Intel (x64)](https://github.com/ggml-org/llama.cpp/releases/download/b11921/llama-b11921-bin-macos-x64.tar.gz) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b11921/llama-b11921-xcfram

SAP advances its autonomous enterprise push with smarter, governed Joule agents

SAP SE today announced a set of artificial intelligence innovations at its annual TechEd conference for developers, enhancing its Joule assistant solutions and Business AI Platform, including tabular data foundation models. New capabilities in Joule Work will help AI agents become more efficient and interoperate with open protocols. In this case, the Model Context Protocol […] The post SAP advances its autonomous enterprise push with smarter, governed Joule agents appeared first on SiliconANGLE.

The climate crisis is becoming increasingly urgent—will AI be a blessing or a curse?

AI’s energy consumption will exacerbate the climate crisis, while climate challenges, in turn, are likely to impact AI’s large-scale development.

Left-hand open-source model, right-hand creative Agent—Hi3D’s AI 3D second act

Author | Cynthia Editor | Zheng Xuan In 2026, the AI 3D landscape stands at a delicate crossroads. Momentum in the primary market continues to surge, with new models and startups emerging one after another, and single-image-to-3D generation now delivers photorealistic results that are almost indistinguishable from real-world objects. Yet beneath the dazzling demo façade, most generated outputs remain stuck at the “turntable showcase” stage: a meticulously crafted character or mechanical model slowly rotates under lighting, its scales, fabric folds, and metallic engravings rendered with lifelike detail. But as soon as these details enter real‑world production workflows—such as game engine rigging, industrial simulation, or 3D printing—their limitations quickly surface: structural fractures, backside collapses, fused details, and uneditable meshes begin to erupt, swallowing up the efficiency gains brought by AI through costly manual post‑processing. Fidelity in frontal views no longer equates to overall 3D quality—a consensus that’s taking hold across the industry. The race for AI‑driven 3D has shifted from the first half—“Can we generate?”—to the second half—“Can we deliver in real‑world applications?” It is precisely at this juncture that leading player Hi3D has made two strategic moves: officially open‑sourcing its production‑proven Hi3D V1.0 platform, while simultaneously launching the creator‑centric Hi3D Agent, which bridges the industry’s foundational layers and deeply embeds itself within the full creative workflow. Prior to this, the company—founded less than three years ago—had already delivered a solid track record: securing two consecutive rounds of funding within six months, totaling nearly $50 million, with top-tier investors like Sequoia China, IDG Capital, and Meituan Longzhu all participating; its Hi3D platform now serves over 120 countries and regions, supporting more than 2 million creators, and enjoys deep integrations with leading hardware manufacturers such as Tuzhu Technology and xTool. From academic breakthroughs in sparse geometry to open‑source, production‑ready models, and end‑to‑end connectivity via its Agent creation system, Hi3D has secured its ticket to the next phase of 3D generation. 01 3D Generation: Overhyped Demos, Underestimated Production Gaps The underlying logic of single‑image‑to‑3D generation inherently carries a perspective bias. As metrics like ULIP and Uni3D have repeatedly set new benchmarks, the industry’s past few years have been marked by an emphasis on front-facing views at the expense of holistic understanding, and on outward appearance rather than internal structure. However, the essence of a 3D asset lies in its ability to be rotated, edited, and manufactured as a three‑dimensional object. In a single character image, the vertical…

b11919

opencl: refine mul mat id -1 index (#31170) Assisted-by: Codex **Website:** - **Attestations:** - **macOS/iOS:** - [macOS Apple Silicon (arm64)](https://github.com/ggml-org/llama.cpp/releases/download/b11919/llama-b11919-bin-macos-arm64.tar.gz) - macOS Apple Silicon (arm64, KleidiAI enabled) [DISABLED](https://github.com/ggml-org/llama.cpp/pull/23780) - [macOS Intel (x64)](https://github.com/ggml-org/llama.cpp/releases/download/b11919/llama-b11919-bin-macos-x64.tar.gz) - [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b11919/llama-b11919-xcframework.zip) **Linux:** - [Ubuntu x64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b11919/llama-b11919-bin-ubuntu-x64.tar.gz) - [Ubuntu arm64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b11