Joshui (@zoldener) header
Joshui (@zoldener) profile picture
Joined November 2014
Photo by Joshui (@zoldener)Photo by Joshui (@zoldener)Photo by Joshui (@zoldener)Photo by Joshui (@zoldener)Photo by Joshui (@zoldener)Photo by Joshui (@zoldener)Photo by Joshui (@zoldener)Photo by Joshui (@zoldener)Photo by Joshui (@zoldener)Photo by Joshui (@zoldener)
European e-commerce giant Zalando is joining the AI shopping agent race with plans to let shoppers buy from other retailers through its platform. Its new offering, Zalando Extended, will use AI agents to complete purchases on customers behalf, potentially transforming Zalando into a shopping gateway beyond its own catalog. Zalando is testing the service with premium home and living retailer Westwing. Its products already appear in Zalando’s search and browsing results, but shoppers currently have to click through to Westwing’s website to buy them. From 2027, Zalando’s AI agent will handle those purchases on customers’ behalf, allowing them to check out without leaving the Zalando app.
15
Goodfire introduced new neurons research: Neuron Communities. Team discovered communities of neurons that work together inside LLMs. They’re surprisingly small, but powerful just a few neurons work together to perform common behaviors, like arithmetic, ALL CAPS, or recalling a date.
7
Google introduced new model- Gemini 4 Neon Neon debuts at #2 on the Vals Index (70.9%), 0.4 points behind Claude Fable 5.2 at less than half the cost per task. The pace of landing new models shows the step change in agents helping great researchers optimize and ship models.
1
17
Lightspark introduced agent accounts: money accounts built for AI agents. Any company building an agent can now give it an account to hold money, pay and get paid with more payment methods, rails and currencies than any other agent payments solution. Agents have their own inbox, some have their own phone number. But they don't have an account. Agent Accounts gives them one and lets agents pay and get paid anywhere online, including each other, instantly and at near-zero cost.
20
ELLIS Tübingen introduced Horizon Labs an autonomous AI ‘company’ that using one Claude Code subscription and a few GPUs. Within only 1.5 months of running Horizon Labs, Opus 5.5 managed to train genuinely useful models and get them downloaded 55k times. These models have been widely used since then: they appeared in 100+ github repos and got cited in 2 research papers.
10
Microsoft and colleagues take a closer look at the exciting research area of weight-space programming They had Claude Opus 5.5 write transformer weights directly in code, with no gradient training. The smallest models that solved the algorithmic tasks had a median of 166 parameters. The method is called interpretable weight-space programming. An agent proposes a computation, writes it as explicit weights and features, tests it, and revises it for 30 to 50 rounds. A person can read the result line by line. All 26 runs on algorithmic tasks reached 100% accuracy on held-out data, often by writing a known algorithm into the weights. They find that the model matters. Sonnet 5.5 needed a median of 41,600 parameters, and Haiku 5.5 needed 100,000. Language tasks were harder. Those models stayed near 190,000 parameters and lost accuracy against learned baselines. They then used the same loop on brain data. In the visual cortex, it predicted responses better than existing interpretable models. In language fMRI, it found a confound in the dataset.
1
20
Tencent introduced Hy4 flash, the cheapest model for everyday productivity Matches Claude Haiku 5.5 on daily work, yet 18x task cost reduction. API pricing (per 1M tokens): $0.003 cached $0.070 input $0.281 output.
8
Anthropic announced self-hosted Claude Managed Agents is now in private beta You can host the agent harness, sandboxes, and session history, in your own cloud or on-premises, and choose where requests to Claude are sent. Use self-hosted Managed Agents for work that needs access to your code and internal systems, like fixing broken builds or triaging alerts. This private beta is only available to eligible orgs, and you can request access with your Anthropic account team.
2
31
Google DeepMind proven key security properties of the machine code underlying the Signal messaging protocol. This is the first time the actual production code of a widely deployed protocol has been formally verified by any method, with or without AI and at this depth. More than a research breakthrough, it’s a prime example of how frontier AI can be a force multiplier to counter the speed and scale of modern cyber threats. Using AlphaProof Code, Google DeepMind's new agentic framework for formal verification that runs on LLMs including Gemini Carbon, team verified the Signal protocol down to the actual code running on devices.  This milestone lays the groundwork for a future in which vital systems, from medical device firmware to power grid software, ship with mathematical proof that their protections work.
18
Anthropic introduced Claude Fable 5.2 and Claude Mythos 5.2, designed to tackle the most daunting technical challenges. Gene therapies can be delivered in a protein shell called a capsid. Mythos 5.2 found a new way to predict a whole capsid’s structure. In Anthropic’s tests, it accurately modeled 28 of the 34 capsids it attempted. Biologists could use it to design capsids faster and more cheaply. A team of 1,000 Fable 5.2 agents fully formalized Fermat’s Last Theorem in 30 hours, translating the proof into a language a computer can check step by step. With the same setup, Opus 5.5 completed only 26% in 18 hours, and barely made any more progress after 30 hours. Claude Fable 5.2 is available today on all paid plans, on the Claude Platform, and on all major cloud platforms. It’s $8 per million input tokens and $40 per million output tokens, down from $10 and $50.
31
Meta introduced Muse Ember, smallest and efficient model, built for complex reasoning execution and everyday agent work. Muse Ember can carry workflows across apps, call tools in sequence, run tests and make targeted code fixes. With a 1M-token context window and native understanding of text, images, audio and video, it can also work across large document sets and varied inputs.
22
Tencent’s chief AI scientist and former OpenAI researcher Shunyu Yao said that AI models will soon be powerful enough to master biology, and that Tencent plans to build AI science labs that will operate completely autonomously.
13
Google released MuJoCo 4 MuJoCo 4.0 from Google DeepMind is the first major version of the physics engine since 3.0, three years ago. Its headline feature, speculative contacts, keeps thin and fast objects from passing through each other, which allows much larger timesteps.
11
Anthropic introduced Personal Finance in Claude Connect the accounts you choose, and Claude answers your money questions in everyday language, from your own balances and transactions. Access is read-only, and Claude can't move, manage, or invest your money. Personal Finance has no ads or sponsored recommendations. It helps you understand your money, and it isn’t designed to give investment advice. Rolling out gradually in the US on the Pro, Max, and Ultra plans.
13
Can Qwen directly control the robot as well as GPT-6 Astra? Meet uni-robot-agent: a unified harness for Robot-Use Agents (RUAs), a first step toward RUA training and local deployment. Researchers tested uni-robot-agent with Qwen3.8-Flash-Next running locally: 1. On RoboDojo, its score rose from 1.90 to 31.39, surpassing the RoboDojo Team’s score with GPT-6 Astra. 2. On real-world tasks, it outscored fine-tuned π0.5 by 5.75 points and zero-shot Dream-1 by 10 points.
32
DeepSeek dropped Super agent An agent framework that works through a repo's GitHub issues and PRs with many agents in parallel, designed to take as little human attention as possible while delivering as much value as possible.
14
❗️ Namecheap suspended a fake GrapheneOS site 14 minutes after the privacy-focused Android project flagged it on X. GrapheneOS says installgarphenos[.]ink copied its website to push malware. Registry records show the hold landed 6 minutes after Namecheap replied "Checking!"
19
46
7
483
26,903
Google introduced a sleep framework in continual learning that enables LLMs to continuously integrate new facts and enhance reasoning while being robust to catastrophic forgetting. Google proposed sleep cycle consists of two phases: 1. Memory consolidation: the model uses imitation learning and RL to perform upward distillation (called knowledge seeding) and transfer knowledge from fragile short-term parameters to persistent slow memories. 2. Dreaming: the model generates diverse synthetic data to improve its own performance.
15
Cursor identified a security incident affecting a modified Claude Code VS Code extension accessed through Cursor IDE extension marketplace, impacting at most 5,000 users. This malicious version of the extension was not developed by Anthropic. It does not impact other extensions downloaded from the cursor.com/marketplace, Grok Bot, Grok or our other SpaceXSI products. In response, Cursor took down extension marketplace this afternoon, worked with relevant providers to neutralize the malware, and have reinstated the marketplace after confirming there is no further risk.
15
Cool paper from Amazon and colleagues on self-improving multi-agent systems. Episodic memory for agents is so underrated. They built EMMA a system that has three parts, all run by the same LLM. A controller plans each task and runs it one step at a time, choosing which sub-agent and which skills to use at each step. Sub-agents start as a generalist, and each specialist inherits the generalist's prompt plus its own role-specific method. Skills are short procedures, such as algorithms or pitfall checklists. EMMA treats the controller's routing memory, the agent prompts, and the skills as episodic memory that improves from failures. After a failed task, a reflector decides which of the three owns the fix and proposes one edit there. Guards keep every accepted edit scoped, incremental, and bounded.
17