All published intelligence, sovereign stack research, ecosystem news, and AI foundation model reviews.
🖥 Tencent rolls out Hy Memory 2.0: open model, framework, and MemHub for long-term AI agent memory. Hunyuan released three things in one release: the Hy-Memory-30B-A3B model, SDK framework, and MemHub application. According to Tencent, this is #1 among open solutions on four...
Человек, который зареверсил продукты Adobe, теперь сделал то же самое с Office. Есть аналоги Word, Excel и PowerPoint, и все они бесплатные. У каждого есть собственный MCP-сервер, а аналог Word даже позволяет Claude редактировать ваши черновики с включённым режимом отслеживания…
I. Meet cute Barnabas and Mainspring grew up near at hand, about an hour apart with good traffic, but for a long time they weren’t close. They traded glimpses and occasional hellos and then went about their business. Both of them carried on livelier conversations with…
✔️ TypeSafe AI added decision-making models Jev vision and long context In the new version of Jev 1.14, the model learned to accept images as input, search for objects on them and work with long documents. All three functions are experimental and their quality is still too...
Big tech companies pose a serious threat to democracy by using artificial intelligence and undermining the quality of journalism. We must unite efforts to protect public interests and counter the practices of AI companies that prioritize profit…
🎮 Epic Games is making a game without “a single line of manual code” According to media reports, the studio is transferring the development of a new shooter with Marvel and Star Wars heroes to AI tools. The team was allegedly tasked with generating code and graphics through Claude Code, leaving the developers primarily to play a role...
Подборка активностей: что отработать сейчас 🟠RUN October Jam — до 2 ноября делаем игру, интерактивную историю или видео с ИИ: 100 000 кредитов на разработку и конкурс с фондом $30 000. Готовый контент можно монетизировать: автору 70 % от подходящего дохода, но вывод денег…
Ведущий инженер Anthropic, написал руководство о том, как действительно ставить более амбициозные задачи перед ИИ, и поделился готовым промптом. Феликс Ризеберг, работающий над Claude в Anthropic, отмечает, что все советуют быть "амбициознее" при использовании ИИ, но никто не…
📌 Google showed how Gemini 4 Argon controls robots and builds simulations for them. The new Argon scored 87% on ERQA, a test of reasoning about the physical world for robots, without external tools, solely through reasoning. According to Google, this is better than any other…
What does it take to build like a $10 billion company? >> The thing that was difficult was okay, we felt like we had this great product, but it's very important how you talk about your product and communicate it. And it wasn't until we changed the tagline to the open source…
Benchmark partners acknowledge that venture capital's traditional rule of not backing rival companies is being tested by the rapid development of AI. Conflicts of interest among portfolio companies are becoming increasingly common as companies quickly...
Claude Code позволяет экспортировать всю сессию с помощью команды /export file_name. Команда создаёт обычный текстовый файл, содержащий всю историю сессии. Его можно использовать в качестве контекста для другого агента, чтобы тот продолжил вашу работу, или проанализировать для…
Okay, this is the craziest thing I've seen in a long time. Elon just showed off the new Roadster, which can like literally fly. The first demo they showed was it actually driving up this vertical ramp, actually going completely upside down, and then leaping off of it, flipping…
arXiv:2610.20665v1 Announce Type: new Abstract: Large language model (LLM) agents can solve complex interactive tasks, yet reasoning and tool calls take time. In continuously changing environments, an otherwise correct action may arrive too late. We introduce RealtimeBench, with…
arXiv:2610.20676v1 Announce Type: new Abstract: Correctness feedback suppresses the very evidence that specification-completeness auditing is designed to uncover: programs that satisfy a written contract but violate intended behavior. We demonstrate this objective reversal on 12…
arXiv:2610.20679v1 Announce Type: new Abstract: Representation-based safety defenses for vision-language models (VLMs) often learn to separate harmful from benign inputs, but separation alone does not establish what a risk score means or when it should be trusted. We introduce…
arXiv:2610.20669v1 Announce Type: new Abstract: Reward-guided discrete diffusion enables test-time alignment with downstream objectives, but applying Sequential Monte Carlo (SMC) guidance at every denoising step is computationally expensive. Active guidance steps require…
arXiv:2610.20675v1 Announce Type: new Abstract: Agents pair reasoning models with harnesses that organize model calls, execute tools, load skill instructions, and retain history in memory. Understanding agent performance requires linking execution conditions to outcomes,…
arXiv:2610.20681v1 Announce Type: new Abstract: Formal inequality proving requires mathematical reasoning and exact computation, often connected through intermediate lemmas and goal reformulations. We introduce IneqAgent, an agentic framework combining LLM reasoning with…
arXiv:2610.20685v1 Announce Type: new Abstract: Symbolic regression searches for an equation that explains observed data. Numerical error is commonly used to guide the search, but it alone can favor an accurate surrogate over a candidate with the right structure. We propose…
arXiv:2610.20805v1 Announce Type: new Abstract: A reasoning trace retained at answer time plays two roles: it can raise accuracy, and it can compete with the current evidence when the two disagree. We disentangle these roles by crossing verified evidence versions with the traces…
arXiv:2610.20786v1 Announce Type: new Abstract: Multi-modal agents have made significant progress in interacting with open worlds. However, their ability to self-evolve remains limited. Supervised fine-tuning enables imitation but not active experience organization, while…
arXiv:2610.20780v1 Announce Type: new Abstract: In LLM-based multi-agent systems (MAS), a local mistake can propagate as other agents build on it, turning a correctable error into a failed task. Diagnosing completed runs or repeatedly re-executing them to search for corrections…
arXiv:2610.20764v1 Announce Type: new Abstract: This paper challenges the understanding capabilities of existing AI paradigms and emerging world models by asking a fundamental question: How quickly can reality change before an intelligent system can no longer reliably perceive,…
arXiv:2610.20782v1 Announce Type: new Abstract: The closed world asks AI for specialized competence and lets it manufacture the iid data its theory assumes: a fixed criterion licenses collecting examples until the task is covered. The open world asks for versatility --- taking…
arXiv:2610.20809v1 Announce Type: new Abstract: A self-improving agent has to decide which new skills to keep, and in practice it keeps whatever earns a high reward, convinces a judge model, or compresses its library, even when the skill is wrong. That is fine until the answers…
arXiv:2610.20836v1 Announce Type: new Abstract: Evaluator-guided reasoning presents a local control problem at every step: trust the current answer, return a fallback, or spend more inference. We formulate this choice as a compute-aware three-action controller, scoring terminal…
arXiv:2610.20845v1 Announce Type: new Abstract: Skill-policy co-evolution, in which an agent's policy model and its skill bank improve each other during training, is a promising paradigm for self-improving agents. Current evaluations score the co-evolved policy on the final bank…
arXiv:2610.20922v1 Announce Type: new Abstract: Compositional structure is pervasive in nature and across scales: atoms combine into molecules, molecules into cells, and cells into organisms, while further compositional hierarchies appear within each level. More broadly,…
arXiv:2610.20926v1 Announce Type: new Abstract: Agent skills turn procedural knowledge into reusable artifacts, allowing LLM agents to accumulate and refine expertise through experience. Building on this, skill evolution iteratively improves skills using execution feedback.…