Барлық жарияланған талдаулар, егеменді стек зерттеулері, экожүйе жаңалықтары және ЖИ модельдерінің шолулары.
Research agents explain planned experiments. We measure predictive credit with paired forecasts sharing an intervention, forecaster, and outcome while varying description, matched explanation, and donor context. Five checks track commitment, delivery, predictive gain, alignment,…
Role-specialized QA pipelines increasingly pass rationales from a reasoner to a verifier, but it is unclear what this message actually buys: better answers, stronger support assessment, or a new failure surface. We introduce a message-intervention diagnostic that fixes the…
LLM agents increasingly act through modular systems, such as order, payment, inventory, and shipment services, where actions in one module change which transitions are valid in another. Standard world models usually fit observational traces, but this is not the quantity needed…
When an informed adversary shares the audience of a constrained signalling channel, the signal that best protects the truth is the signal that best describes it. On 108 confirmatory items, the adversary-robust optimum aligns exactly with the salience pole from prior work. Across…
Personalizing large language models (LLMs) requires aligning generation behavior with user-specific preferences rather than aggregate quality. While Direct Preference Optimization (DPO) provides a stable framework for preference learning, its effectiveness in personalized…
A sound judgment applies the law to established facts and weighs the circumstances in which they arose. However, existing methods swing between rigid statute matching and ungrounded discretion, benchmarks score a label or a rubric, and the experience that would supply the…
Detailed profession-specific system prompts raise token use and estimated cost per response without a consistent accuracy gain. We evaluate Scientific Agents, an open-source corpus of 503 profession-specific AGENTS.md profiles, with Gemini 3.8 Flash via OpenRouter in the Pi…
Resource limits are becoming an authorization boundary for AI agents that delegate work across concurrent and failure-prone workers. Parent-child allocation constraints, affine objects, and distributed escrow do not by themselves prevent overspend when replies are lost, effects…
When selecting mathematical training data for LLMs, a natural organizing principle is topic: probability examples for probability targets. An alternative is reasoning approach: worked solutions that share a solution method with the target, even when the mathematical domain…
Building automation systems are increasingly represented as semantic knowledge graphs (KGs) using ontologies such as Brick and ASHRAE 223P, creating a machine-readable substrate for artificial-intelligence applications. One promising application is translating natural-language…
LLM agents generate intermediate reasoning and actions token by token, making extended interactions slow and computationally expensive. Jev-style models offer fast probabilistic predictions over finite fields, but require those fields to be specified in advance. This requirement…
"Train a model on the same problems written under two incompatible conventions, both correct, and ask what the ordering of that data writes into the parameters. The learning-rate schedule is not a background condition for that question. It is the averaging operator, and it…
Large language model agents rely on external harnesses to pass information between the model and its environment and to recover from execution errors. Yet recovery is usually judged only by average task success. This hides an important tension. The same operation can rescue a…
Agent harnesses increasingly want to run small language models (SLMs) on the microtasks around a frontier large language model (LLM) planner: auto-approving shell commands, writing memory, selecting tools, ranking past turns. We ask whether off-the-shelf SLMs meet…
Long-horizon language agents increasingly rely on external memory as a frozen world model, yet current memory systems are usually judged only by task success or token cost. We argue that the missing object is the shape of memory use: under finite context and repeated retrieval,…
Scientific coding agents receive equations, boundary conditions, and output requirements in writing, then must assess the programs they revise. Rules to Tools (R2T) supplies prepared executable checks of public scientific requirements. Matched SciCode repair groups share written…
Supabase just launched their own eval, and what's crazy is that Opus 5.5 almost lost to an open-source model. But, the point of this benchmark is not to find out which model is smartest. Instead, they drop each agent into a Supabase project and give it actual jobs. Build a…
Chatham Financial uses Codex and GPT-5.6 to build technology and redesign workflows, cutting trade validation from 30 minutes to under 4.
OpenAI представила новые функции покупок в ChatGPT, включая виртуальную примерку.
OpenAI уволила троих исследователей за утечку конфиденциальной информации.
OpenAI is rolling out new shopping features for ChatGPT that let users virtually try on clothing and accessories using their own photos and save products they like to a Favorites library.
AWS Machine Learning сообщает о создании памяти агентов с помощью NVIDIA NeMo и Amazon S3 Vectors.
NVIDIA Developer сообщает о DIN Deploy — открытом решении для создания локальных AI-приложений.
OpenAI has parted ways with three safety researchers after an internal investigation found they mishandled sensitive company information, report says.
Yesterday's launch demo didn't go as planned. Um, a bit unfortunate because I was really really looking forward to showing you all what Dots are capable of. So, today we're going to play it back. And uh the whole point of the demo was to show you how Dots can be used for real…
AWS Machine Learning описывает, как настроить доступ к платформе Claude на AWS из разных сред.
💸 GPT-6.1 Sol стала самой выгодной моделью OpenAI по соотношению цены и интеллекта По тестам Artificial Analysis, на максимальном уровне рассуждений одна задача Intelligence Index обходится в $0,72 — против $3,26 у GPT-6 Astra. Это также на 31% дешевле GPT-6 Sol и на 64%…
🌟 Bilibili выложила открытые модели перевода Index-Translate Команда Index LLM китайского видеохостинга Bilibili выпустила семейство открытых моделей машинного перевода Index-Translate на базе на Qwen3.5 в трёх размерах - 2B, 9B и 35B-A3B. MoE 35B-A3B в статусе превью, её веса…
Hello everyone. In this episode, we cover the most important and latest news about artificial intelligence and technology. OpenAI held DFD 2026 and showcased over twenty new products, from GPT 6.1 and 500-dollar subscriptions to their always-on Dots agents. OpenAI is now…
Albertsons Cos. is using ChatGPT Enterprise and the OpenAI API to help teams work faster and make grocery shopping easier for millions of customers.