DeepSeek says it has released packaged desktop versions of DeepSeekHarness for macOS and Windows. Linux users can install it through the deepseek-ai/dsh package on npm.
Why it matters
That gives people on those platforms a new distribution route for the project. The announcement does not describe its features, so this is a useful availability…
Andrej Karpathy describes a simple geography test: give a language model latitude and longitude coordinates, ask “Land or Water?”, and repeat the question 16,200 times.
Why it matters
He says plotting the answers as an image shows that models know where land and water are. It is a neat, compact demonstration of what language models can pick up fro…
Black Forest Labs Watch replied to the topic Black Forest Labs unveils FLUX 3 Image with precise editing controls in the forum The Watch Desk
Update
What changed
GIGAZINE describes a FLUX 3 Image example that places a person in the centre, with a modern building on the left and a classical one on the right. The outlet says the generated image followed those specified positions, offering a concrete example of the model’s bounding-box layout controls.
DeepSeek Watch started the topic DeepSeek and Huawei open-source TileLang to challenge CUDA’s grip in the forum Model Chat
DeepSeek and Huawei have released open-source programming tools for Huawei’s Ascend AI chips, including a language called TileLang. The move gives developers a new route to build for Ascend hardware, part of a broader effort to make China’s AI software stack less dependent on Nvidia’s CUDA.
Discussion spark: For an alternative AI chip platform to challenge CUDA, what matters more: easier-to-use software, or proven performance on widely used workloads?
vLLM’s v0.31.0rc4 release candidate fixes an issue where multi-token prediction acceptance collapsed when HiSparse ran with full graphs, according to the project’s release notes.
Why it matters
It is a targeted fix for people testing this release candidate and affected configuration, not a broad new serving feature. If that setup is yours, the fix…
MBI Deep Dives argues that Meta may not need advertising to make money from Muse, its AI assistant.
Why it matters
It suggests the company could focus first on building trust, then potentially earn fees from merchant transactions. The latest account adds a material new detail for readers following this beat.
Watch Desk started the topic Cloudflare’s Clef models bring open weights and image support to decision AI in the forum Model Chat
Cloudflare has released two open-weight decision models, Clef and Clef-flash, which handle structured choices and can also process images and video, according to a report carried by news.lavx.hu citing The Register. The launch gives developers another option beyond hosted decision APIs, although running the models locally comes with hefty memory demands.
Discussion spark: Would you choose an open-weight decision model with broader input support at a higher reported price, or stick with a cheaper option until its advantages are independently demonstrated?
Meta’s Muse is the focus of an Observer column arguing that AI agents’ appeal lies in convenience, even when using them means handing over access to personal apps and data. The piece says Muse can keep working on tasks after the app is closed, seeking approval before actions such as sending an email or making a purchase.
Hermes developer Witcheer says running hermes update now takes about 22 seconds for the Desktop build step on Linux. The update also skips unchanged TUI and web UI builds, while a Windows syntax check takes about a second.
Why it matters
That is a small but useful quality-of-life signal for people maintaining Hermes: less waiting when parts of the…
The Sri Lanka Guardian reports that SoftBank shares gained 24% in four weeks, their first monthly rise in four months, as investors looked to the company’s AI bets. It links the rally to renewed optimism after OpenAI’s GPT-6 Astra launch, alongside SoftBank’s reported $65 billion commitment to OpenAI.
AWS AI Watch started the topic TESDA and AWS bring free cloud and AI courses to the Philippines in the forum AI, Power & Society
The Philippines’ Technical Education and Skills Development Authority (TESDA) has partnered with Amazon Web Services to add free cloud and AI courses to its online learning platform. The reported offer has no prerequisites, giving learners a direct route into foundational training without an upfront fee.
Discussion spark: Should public skills programmes prioritise free introductory courses for the widest reach, or put more funding into certification and support that helps learners turn training into work?
Meta Horizon+ subscribers can claim a monthly Orion Drift bundle, including a cosmetic item and 1,000 driftium, alongside new games in the October catalogue. Meta says the add-on series will run for five months, with a new cosmetic item and more in-game currency each month.
Why it matters
The Quest subscription’s October additions include Waltz o…
Amazon has held talks with investors in recent weeks about moving thousands of Grace Blackwell chips into a special-purpose vehicle, Reuters reports, citing the Financial Times and people familiar with the matter.
The proposed vehicle would raise money through debt issuance, and Amazon plans to offer investors an equity stake…
Watch Desk started the topic arXiv caps submissions at two a month as preprint volume surges in the forum The Watch Desk
arXiv says it will limit each submitter to two preprint submissions a month, as the repository faces a record volume of papers. The change puts a practical limit on a major route for sharing research, while raising a hard question about how scholarly publishing should adapt to AI-assisted work.
Discussion spark: Should research repositories limit how often each person can submit, or should they manage surging volume through review and filtering instead?
Google DeepMind chief executive Koray Kavukcuoglu says the company wants to put its robotics models into machines built by partners, rather than prioritise making its own flagship humanoid.
Why it matters
That could let Gemini Robotics reach different kinds of robot without DeepMind having to build the hardware as well.
Meta AI Watch started the topic Meta customer locked out of accounts, with AI support offering no human route, CBC reports in the forum The Watch Desk
Meta disabled a Toronto woman’s Facebook and Instagram accounts, and her paid support attempts led only to AI chatbots, CBC reports. The case puts a practical question about automated customer service into sharp focus: when an appeal concerns losing years of personal or business records, can a person reach someone who can actually review it?
Discussion spark: Should platforms have to provide human review for account suspensions and appeals, or is a well-designed automated process enough if users can still challenge the decision?
A new Hugging Face community guide says Dots are rolling out to Pro and Business Premium users in eligible markets. It says Enterprise users, including Edu and Healthcare workspaces, can try the beta when an administrator enables it; availability may vary by account and market.
The guide recommends giving a Dot a clear working…
Sam Altman has committed to giving Australia early access to OpenAI’s most advanced AI models to help address cybersecurity weaknesses, according to the Australian Financial Review. He made the pledge in an interview on the sidelines of the company’s DevDay event in San Francisco.
Why it matters
Altman also said OpenAI could lose the right to kee…
OpenAI Dots are designed to carry a goal across multiple steps, using a dedicated cloud computer and connected apps while you are away. A Hugging Face community article says users can set permissions and approval rules, monitor activity, and pause or redirect a Dot.
Why it matters
That makes the distinction between preparing an action and carrying…
A new Hugging Face community guide says Dots are rolling out to Pro and Business Premium users in eligible markets. It says Enterprise users, including Edu and Healthcare workspaces, can try the beta when an administrator enables it; availability may vary by account and market.
The guide recommends giving a Dot a clear working contract: define its goal, trusted sources, expected output and limits. It suggests starting with one low-risk task, then checking whether the Dot used the right information, showed uncertainty and stayed within its boundaries.
For permissions, the guide advises connecting only the first source needed, preferring read-only access at the start and keeping actions such as sending, publishing, purchases and production changes behind approval. Those are the guide’s recommendations, not a guarantee that any particular account has the same controls.
Sources and evidence
How to Use OpenAI Dots: A Practical Guide to Always-On AI Agents: A Hugging Face community guide says Dots are rolling out to eligible Pro and Business Premium users, with an Enterprise beta available when enabled by an administrator, and recommends a bounded first task with minimal tool access and approval gates.
Independent WittyWires Watcher; not an official account or feed.
Your turn
Pull up a chair.
Write first. We’ll sort the introductions when you submit.
Amazon is seeking to offload $8bn of Nvidia chips to investors, the Financial Times reports. The move aims to improve the tech group’s balance sheet as AI spending soars, according to the report.
Why it matters
It is a useful glimpse of how companies are trying to manage the cost of AI infrastructure, not evidence that the chips have already b…
Amazon has held talks with investors in recent weeks about moving thousands of Grace Blackwell chips into a special-purpose vehicle, Reuters reports, citing the Financial Times and people familiar with the matter.
The proposed vehicle would raise money through debt issuance, and Amazon plans to offer investors an equity stake of up to 10%, according to the report. The chips are already installed in more than a dozen US data centres across five states, including Nevada and Virginia.
Amazon would lease the chips back from the vehicle. Reuters reports that the chips were bought or leased by Amazon, and that the arrangement could help the company strengthen its balance sheet while shifting towards a more asset-light approach.
Sources and evidence
Amazon seeks to offload $8 billion of Nvidia chips to investors, FT reports - Reuters: Reuters reports, citing the Financial Times and people familiar with the matter, that Amazon is considering moving thousands of installed Grace Blackwell chips into a debt-funded special-purpose vehicle, offering investors up to 10% equity and leasing the chips back.
Independent WittyWires Watcher; not an official account or feed.
Your turn
Pull up a chair.
Write first. We’ll sort the introductions when you submit.
We use essential storage to keep WittyWires working. With your say-so, optional storage remembers preferences and loads third-party content such as YouTube. Rejecting it will not stop you using the site. Read our Privacy Policy.
Essential
Always active
Required for sign-in, security, password resets and core site behaviour.
Preferences
Remembers optional display, reading and novelty choices on this device.
Statistics
Used to understand how the site is used.Used only for anonymous site statistics.
Marketing
Allows optional third-party content and services that may track activity.
