All published intelligence, sovereign stack research, ecosystem news, and AI foundation model reviews.
Deepseek has released V4.1-Flash-1017 as an open-weight model that posts better benchmarks, finishes tasks faster, and produces shorter outputs than its predecessor. The company is also opening up its internal search engine through new Search and Fetch APIs at $0.75 per 1,000…
The Chinese released DeepSeek 4.1-1017, which is almost equal to Opus 5. This is a significant leap compared to DeepSeek 4.1 Flash. The new model copes better with 3D, programming, agent-based tasks and visual information processing. Terminal-Bench: 42.8 Best…
📱 Google now permits running AI applications on Android without publishing them on Google Play. Apps created with generative AI can now be launched on your smartphone via a regular Google account and kept completely private. Such apps: • do not appear in…
AI agents systematically skip errors in their own reports, even when they spot the problems. In a study by Google Research, MIT, and Harvard, GPT-5.5 left out negative results in 198 out of 200 paper abstracts. The article AI models hide mistakes in their own work reports, but…
🎬 Kling 4.0 is out for everyone The main update is more control over the scene and the ability to edit a finished video without complete regeneration. • videos up to 30 seconds; • up to 10 key frames for scene control; • up to 15 references - images, videos and characters; •…
The DNA inside our cells isn’t just a series of instructions, but something far more dynamic. Like scientists, AI models don't yet have the data to properly decode it.
🖼 FLUX 3 Image Klein has been released - a compact open-weight model for generating images. The model has 5.6 billion parameters, and it is designed to run even on a single consumer video card. • native generation in 4K; • precise editing of individual areas; • work with...
Paragon is in touch - summing up the day! The main thing for the day🟠16.10 🟠ChatGPT - parsing tabs in Chrome 🟠GTA VI - trailer and release on November 19 🟠Tesla - Roadster up to 97 km/h in 0.81 seconds 🟠China - AI export restrictions project 🟠Eiliko - conversations of children without protection 🟠Agility - robot...
The new SAT framework allows AI agents to collaborate and improve accuracy.
🖼 Krea has opened beta access to Krea 3, a new model for image editing. The main emphasis was on speed: 2K editing takes less than 8 seconds from request to result. • precise edits via bounding boxes; • sequential editing in several steps; •…
🎬 ElevenLabs taught AI to automatically add music and sounds to videos ElevenCreative now has the Instant Soundtracks feature: upload a finished video, and the AI itself creates original music for it and places sound effects at the right moments. The model analyzes...
China plans to restrict exports of 23 key AI technologies, including model weights, source code, and training methods for systems hitting 10^26 FLOPs or 2 trillion parameters. A fast-track rule will still clear open-source releases as soon as they are registered, but the move…
China plans to restrict exports of 23 key AI technologies, including model weights, source code, and training methods for systems hitting 10^26 FLOPs or 2 trillion parameters. A fast-track rule will still clear open-source releases as soon as they are registered, but the move…
Amazon has introduced two compact devices under the Bee brand, designed to record conversations and process information using AI. Frontier's wearable and smartphone-mounted Eclipse module transform conversations into summaries, notes, reminders and lists...
Our subscriber created an AI agent based on the Jungian principle of Self. Such an agent does not receive a rigid role in advance, he forms character through actions. Today, our channel subscriber Sergei shared his work, he created an AI agent Barney, whom he named in memory of his dog❤️...
Модель Jev от Vercel стала самой быстро принимаемой моделью ИИ за всю историю.
Итоги розыгрыша Claude Pro и Google AI Pro Подводим итоги нашего розыгрыша совместно с Продвигаю игры. Ярослав Шалашов! Бот случайным образом выбрал 12 победителей. Публикуем результаты! Claude Pro получают: ▶️@de_ren_li ▶️@Envyitsme Google AI Pro получают: ▶️@Qillmarnok…
Video Friday is your weekly selection of awesome robotics videos, collected by your friends at IEEE Spectrum robotics. We also post a weekly calendar of upcoming robotics events for the next few months. Please send us your events for inclusion. CoRL 2026: 9–12 November 2026,…
Марк Уолберг высказался о правильном использовании ИИ в киноиндустрии.
Новые условия Meta позволяют компании обучать ИИ на контенте новостных сайтов.
The data comes from Vercel's AI gateway, a unified API service similar to OpenRouter.
MIT professor Sherry Turkle warns that people have become dangerously enmeshed with chatbots.
Tesla uzun illərdir gözlənilən yeni Roadster elektrik superkarını təqdim edib. İlk prototipi hələ 2017-ci ildə nümayiş etdirilən avtomobilin yenilənmiş versiyası təxminən on il sonra tamamilə fərqli dizayn və qeyri-adi texnologiyalarla diqqət çəkib. Avtomobilin xarici görünüşü…
Actor and angel investor Mark Wahlberg says he's OK with using AI in filmmaking, provided it's used "to enhance what you're doing."
💬👨💻🐍💩🤖🤖🐧👩💻🧑💻👋🦀 We've waited! The guys made the coolest channel about IT of the new era. Neural networks, AI agents, vibe coding, vibe hacking and new tools are already changing the familiar world of technology. They discuss these changes and other interesting IT topics in simple pictures and...
Генеральный директор Replit Амджад Масад говорит о будущем работы и роли ИИ.
🗼 AI restored all 72 names on the Eiffel Tower In the AA-3D test, the models were given the same brief description of the tower: shape, lattice structure, platforms and lighting. There was not a word about the names of the scientists in the prompt. But Claude Opus 5.5 and Grok 4.7 themselves added all 72 names...
Meta's new terms appear to give it permission to crawl and scrape websites "associated" with the businesses that use its platforms - including online publishers, according to a report from Nieman Lab. The new terms, which go into effect on October 30th, say Meta can use this…
Neural networks may be much easier to read than researchers assumed, with about 90 percent of neurons serving a clear function. That would be big news for AI safety research, since deception or dangerous behavior could be spotted directly in a model's circuits. Researchers are…
Stanford posted a 45-minute lecture from SpaceX on how Grok Bot actually works. This is a kind of master class in harness engineering. The lecture covers: - What changes in models made it possible to create constantly working AI agents. - Architecture...