Bambu Lab says its first 3D scanner, the S1, will go on sale in November, marking the company’s entry into the 3D scanning category. The article says AI is beginning to automate post-scan data repair and reconstruction, while current products can be expensive and complex to use. Industry observers say the S1 could simplify the scanning workflow while retaining professional-level capabilities.
ByteDance Seed announced in an internal email that it will establish the “Seed Model-Alignment&Preparedness” department, led by Yan Lin and reporting to Seed head Wu Yonghui. The original Responsible AI team and some Chat team members will move to the new department. Seed says it will work on model standards, safety evaluations, alignment training, safeguards, and risk monitoring, and collaborate publicly with external research institutions and third-party evaluators.
art cannot be created or destroyed, only remixed (pure code recreation of *the video* using muse wonder)
Qwen has opened applications in the Qwen app to try its embedded personal agent, Qwen Go!, with access expanding gradually to more users. A reporter found it could handle flight delay claims, venue and performance reservations, restaurant bookings, product price comparisons and purchases, and job applications. In one user's test video, it checked a flight delay and contacted platform customer service to submit a claim, resulting in 70 yuan in compensation.
NVIDIA announced planned investments in Doctolib, Entalpic, Genesis AI and Pigment, which are working across healthcare, materials discovery, physical AI and business planning. The announcement adds to NVIDIA’s previous investments in French technology companies Alice&Bob, Gradium and Scintil Photonics. Across Europe, NVIDIA has committed nearly $10 billion to the technology ecosystem.
Qianwen’s Personal Agent, “Qianwen Go,” is coming, and a user said it helped secure a 52-yuan Taobao price protection refund. Another user reported receiving 70 yuan in flight delay compensation after applying through Qianwen Go. The assistant can also handle tasks such as car maintenance, sports bookings and price comparisons.
On October 13, Alibaba’s 1688 announced its integration with Accio, a global AI e-commerce workspace, with immediate access for over 100 million annual active buyers and over 1 million manufacturers. Ahead of Double 11, Accio offers small buyers a procurement assistant for global sourcing, price comparisons, and negotiations, plus an e-commerce manager connecting Taobao, Tmall, Douyin, Pinduoduo, and JD.com. Manufacturers gain a 1688 business assistant with differentiated AI capabilities for in-stock, custom, and cross-border business scenarios.
1688 announced on October 13 that its more than 100 million annual active buyers and over 1 million source suppliers can now activate Accio, its global AI workspace for e-commerce. Buyers can log in with their 1688 accounts to access procurement and store operations tools; sellers can download the 1688 Business Assistant in Accio and link their stores. The assistant supports product listings, advertising analysis, customer inquiries, and management across multiple stores, with capabilities tailored to in-stock sellers, custom manufacturers, and cross-border merchants.
vLLM praises ReplaySSM work by @zwljohnny, @tri_dao, and @NVIDIA, saying Mamba2 standard decode and FlashInfer-backed MTP support are already in vLLM. Caching SSM inputs and updating the checkpoint only when needed reduces recurrent-state memory traffic. In the quoted post, @zwljohnny reports production use in GLM-5.3-Flash and Kimi-K3, availability in TensorRT-LLM, vLLM, FlashInfer, and SGLang, speedups of up to 1.47x for standard decoding and 2.1x for speculative decoding, and low-precision SSM state whose rounding error never propagates through the recurrence.
GPT-4 was the undisputed best model in the world for a full year and we all debated whether anyone could catch up or if OpenAI had some secret formula that nobody else knew about. Now there are are 19 AI models that can run on a modern smartphone that are better than GPT-4!
Baoyu summarizes Anthropic research in which GLM-5.2 alone completed a replication workflow across simulated Android servers, while smaller models failed. The experiment used unauthenticated ADB access and ran GLM-5.2 remotely to deploy Llama3.1-8B, limiting what the results establish about autonomous spread. Anthropic roughly projects that smaller models could match GLM-5.2’s tested capability by year-end based on general capability trends. Researchers also manually distributed Gemma4-26B-A4B across 15 Android TV boxes, achieving 0.39 tokens per second.
Why it matters: The experimental setup helps distinguish successful replication under simplified conditions from autonomous spread on consumer devices, clarifying the limits of the projected timeline for smaller models.
I've been a personal supporter (through our family foundation) of @Khipu_AI & @DeepIndaba, which are fantastic events for AI researchers in Africa and South America. I've encouraged Google to support these in the past. Discovery Loop (@DiscoLoopAI) is thrilled to support!
Great work from the @IntelTech team bringing vLLM support to the new Xe3P-based Intel GPU D-series. More hardware choice for serving open models, with vLLM and the open PyTorch ecosystem.
Gemini
@lillyguisnet posted: “New sota for my image pointing.” Alexandr Wang: now this is the kind of benchmarking i can get behind
Codex resets today. The reset will be completed by the end of the day Pacific Standard Time.
Amazing to see research move from the lab to real-world clinical utility—congratulations to the team on this milestone and The Lancet paper! https://twitter.com/sundarpichai/status/2108329198955958339 In the quoted post, Sundar Pichai describes Google’s prospective AMIE study with @BIDMC_Medicine involving real urgent care patients. Clinicians reported that its medical-history summaries helped prepare for visits in 75% of cases and influenced care approaches in more than half, while its differential diagnoses matched doctors’ final diagnoses 90% of the time.
Baoyu explains that existing Cursor subscribers can keep their plans or switch to The X Subscription, which bundles Cursor, Grok, and X Premium. The new plans exclude third-party model allowances and charge separately at API prices, while existing plans include some usage. Baoyu recommends keeping existing plans for Claude, GPT, or Gemini coding workflows and considering a switch for Grok users; how long existing plans remain available and whether switching back is possible are unspecified.
Alexandr Wang says the issue causing unusually fast quota usage for muse wonder and muse wonder full has been identified and fixed. Users will now see much lower usage against their limits. Wang also announces a muse code reset, as promised after the fix. Baoyu comments that Muse now has enough users that he feels awkward saying Meta also reset limits in an unnoticed corner. 😅
fill your soul with wonder
@doe_naps shared a “decent output” from Muse Wonder after prompting it to create an interactive Three.js experience inspired by Olafur Eliasson’s exploration of natural phenomena. Alexandr Wang reacted: “that's pretty cool!”
i been musing all day
Anthropic says it has made significant progress over the last few months in reducing the effectiveness of distillation attacks against Claude. The company explains how in this post: https://www.anthropic.com/research/distillation-defenses
My AI agent doing proactive customer service: Someone emailed me that, after subscribing to my Substack, they were automatically signed up for others (which shouldn't happen) Without my asking, my Dot ran its own test & figured out what happened by trying to replicate the error.
Artificial Analysis launched AA-3D, a benchmark in which AI agents turn 55 written briefs across seven domains into interactive three.js scenes for blind human comparisons. Over 10,000 counted votes from over 1,000 panelists rank 21 models by Elo with 95% confidence intervals; Claude Sonnet 5.5, Claude Opus 5.5, GPT-6 Astra, and Claude Fable 5.1 are statistically tied at the top. Models receive up to three hours and 250 turns, and reported costs range from $0.20 to $32.08 per scene.
we are fixing issues as fast as we can see them! there's definitely still issues we're trying to iron out, but thank you to everyone for giving us a bit of time to work through all the kinks 🫶
this is a work of art holy shit
People take the Artificial Analysis Index and its individual scores way too seriously.
Cursor is now part of The X Subscription. Current Cursor subscribers can keep their plan or switch to The X Subscription. Paid plans share one usage pool across Cursor, Grok Build, and other Grok products, with twice the coding usage on Grok models as the equivalent Cursor https://t.co/PwXLcH6zHV In the quoted announcement, @SpaceXSI says plans across Grok, Cursor, and X Premium have been unified under The X Subscription. https://t.co/SvkrVvBa0p
this was a big focus area for muse wonder; our goal was to build a model for everyone!
Muse Wonder as a video director
@tobi said, “This model is very very good. I was struggling to find anything to throw at it that it couldn’t do, even when other top models failed. Meta is back.” Alexandr Wang said, “I was struggling to find anything to throw at it that it couldn't do, even when other top models failed.”
expect the unexpected
@SPAC89 praised Meta’s Muse Wonder Full 1.0 Max, saying it produced high-quality 3D results in 22 minutes while using 5% of their weekly limit. They compared it with Opus 5.5 Xhigh and GPT-6.1 Sol Ultra, and called Muse Wonder a frontier model among the top 1–2 models. 🍉 MUSE WONDER 🍉
we have identified and fixed the issue that was causing the unusually fast quota usage for muse wonder and muse wonder full. you will now see much lower usage against your limits. also, as promised, now that we've fixed the issue... muse code reset!!
@bubbleboi says Muse Wonder Full is one-shotting a full RISC-V 64-bit core with a floating-point and SIMD unit, plus extensions for hardware acceleration of the Ed25519/secp256k hashing algorithms. They say they have it working on OpenROAD, which it downloaded on its own, and will tape out later this week. Alexandr Wang: use muse wonder full to make some chips!
Epic Games is reportedly shifting its Disney collaboration to AI-first development, with models writing all code and guiding artwork for an upcoming shooter, according to Bloomberg’s unnamed sources and internal employee messages. Human developers would manage generated assets using models such as Anthropic’s Claude; the $1.5 billion project’s shooter, slated for March 2027, has reportedly been delayed indefinitely amid the shift and understaffed, slower-than-expected development. Epic confirmed it is embracing AI-accelerated development, while CEO Tim Sweeney said the aim is to help employees build more and better work.
Yes, and there is a case to be made that OpenAI and Anthropic will do the best job of figuring out how to optimize AI cost/intelligence with combinations of agents of different ability levels that have been specifically trained to work well together in their own harnesses.
Nathan Lambert calls a piece on China’s LLM data economy the best he has read so far on how it compares with the US. He says Chinese labs seemed 6–12 months behind US labs in willingness to spend on data, but are rapidly catching up, and notes that Tencent, Alibaba, and ByteDance lead Chinese spending rather than OpenAI and Anthropic.
Muse wonders of the world
