RUСША приказали федеральным агентствам удалить AI DeepSeek из систем
США приказали удалить технологии DeepSeek из государственных систем.
SMNTCN (Semanticon) — глобальные новости ИИ
Tag / Entity•Company
#DeepSeek
All stories, analysis, and breaking news for this tag (158)
US orders federal agencies to remove China’s DeepSeek AI from their systems
The US has ordered federal civilian agencies to remove DeepSeek products and third-party services built using its technology, citing national security concerns.
TechCrunch — AI
InfrastructureStory: DeepSeek открыла исходный код DeepInfer - собственного движка для запуска ИИ-моделей.#Nvidia#DeepSeek1 min
DeepSeek has open-sourced DeepInfer, the inference engine it is using to launch V4.1-Flash at scale.
DeepSeek has open-sourced DeepInfer, the inference engine it is using to launch V4.1-Flash at scale. https://github.com/deepseek-ai/DeepInfer It is this code that helps to quickly service requests to models and reduce the cost of their operation. For companies…
Telegram — Вайб-кодинг
The Download: the claim shaking up AI and the internet’s agent problem
This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. “We were looking at it wrong”: The explosive claim that’s upended the AI world “I think I have good news about the nature of neural…
MIT Technology Review
US bans Deepseek technology across federal civilian agencies
The US Department of Homeland Security has banned federal civilian agencies from using AI technology from Chinese lab Deepseek. The article US bans Deepseek technology across federal civilian agencies appeared first on The Decoder.
The Decoder
Anthropic trains agent negotiators
✔️ Anthropic trains agent negotiators Agents are taught to negotiate on behalf of the user, and the company transfers the practices of professional negotiators into models in consultation with law firms. Researchers are interested in the protocols of interaction between agents and...
Telegram — AI Machine Learning Big Data
InfrastructureStory: DeepSeek открыла исходный код DeepInfer - собственного движка для запуска ИИ-моделей.#GPU & Суперкомпьютеры#DeepSeek1 min
DeepSeek has opened the source code of DeepInfer, its own engine for running AI models.
DeepSeek has opened the source code of DeepInfer, its own engine for running AI models. It is this that ensures DeepSeek-V4.1-Flash operates with high performance and minimal latency. The main thing: - Speed up inference - more requests in less time. - Effective...
Telegram — AI Machine Learning Big Data
RUMiniMax выпустила M3.1 Flash — новую недорогую модель для кодинга
⚡️ MiniMax выпустила M3.1 Flash — новую недорогую модель для кодинга Ранее модель можно было бесплатно протестировать в OpenCode под кодовым названием Space Bunny. M3.1 Flash получила контекстное окно на 1 млн токенов и поддержку текста, изображений и видео. По внутренним тестам…
Telegram — GPT Time | Нейросети | ИИ
Datacenter at Home - GB300 GPU - w/ Asus DGX Station (ET900N G3)
What is going on everybody and welcome to a very exciting review of ASUS's take on the Nvidia DGX station. They're calling this the ET900N G3, but I'm going to keep calling it the DGX station or ASUS station for short. What we have here is 748 GB of coherent memory. That's kind…
YouTube — sentdex
The Chinese guy showed what he is capable of: he single-handedly used AI agents to attack 1,200 websites.
The Chinese guy showed what he is capable of: he single-handedly used AI agents to attack 1,200 websites. Today, Reuters reported on yet another Chinese-speaking hacker who, using AI agents, tried to attack more than 1,200 sites and successfully hacked sites associated with more than...
Telegram — Вайб-кодинг
Training-Free Prompt Selection for Efficient Reinforcement Learning With Partial Rollouts
arXiv:2610.24229v1 Announce Type: new Abstract: Reinforcement learning with verifiable rewards (RLVR) has emerged as a prominent paradigm for training reasoning models, but its computational cost makes efficient data selection increasingly important. Since training examples…
arXiv — cs.AI
RUМодель M3.1 Flash от MiniMax обходит конкурентов по цене, но с высоким уровнем ошибок
MiniMax представила модель M3.1 Flash, которая снижает затраты и конкурирует с DeepSeek.
SMNTCN (Semanticon) — глобальные новости ИИ
RUTenstorrent достигает оценки в $6,2 миллиарда для масштабирования ИИ-чипов
Tenstorrent привлек $1,5 миллиарда при оценке в $6,2 миллиарда для масштабирования своих ИИ-чипов.
SMNTCN (Semanticon) — глобальные новости ИИ
MiniMax M3.1 Flash shows China's AI price race has no floor in sight
MiniMax's M3.1 Flash jumps 13 points over its predecessor to 42 on the Artificial Analysis Intelligence Index, at half the cost. It also hits 73.7 percent on the DeepSWE v1.1 coding benchmark for under $0.30 per task, matching Deepseek V4.1 Flash 1017 on both performance and…
The Decoder
RUСША запрещают использование DeepSeek AI в правительстве на фоне напряженности в технологиях с Китаем
США запретили использование технологий DeepSeek AI китайской компании в большинстве государственных систем из-за соображений безопасности. Министерство внутренней безопасности запретило использование DeepSeek в гражданских агентствах, ссылаясь на потенциальные риски, связанные с…
Digital Donut AI Digest
Four AI models recreated Paris 1996 using code - the difference in price was huge
🇫🇷 Four AI models recreated Paris 1996 using code - the price difference turned out to be huge The developer gave four neural networks the same prompt: to create the atmosphere of Paris 1996 entirely using programming, without additional attempts. Results: • Claude…
Telegram — GPT Time | Нейросети | ИИ
The race for the smartest AI model no longer brings money
The race for the smartest AI model is no longer making money. Over the past few months, OpenAI, Anthropic and Google have released more than a dozen new models, a pace that was not seen last year, it should be noted, and it will only grow. At the same time, before a model, if it was 10 points...
Telegram — blockchainRF (блокчейн / мозг / space / WEB 3.0)
DeepSeek has undergone a reorganization + released an update to the Flash model
DeepSeek has undergone a reorganization + released an update to the Flash model. The company has combined its Harness, Web, App and Platform teams into one product department under the leadership of Cui Tianyi, who recently led only Harness. This whole reorganization comes down to the logic that...
Telegram — blockchainRF (блокчейн / мозг / space / WEB 3.0)
How DeepSeek Uses Less Memory
the supply is constrained and thus the prices rise. So going back to DeepSeek, the KV cache only needs a fourth of the HBM and then they also need 1/8 of the SSD storage. Well, they're able to serve many more tokens without necessarily having to pay for all that memory. The…
YouTube — Matthew Berman
GPT-6 Astra solved 98% of complex physics problems and found errors in scientists' answers
🧠 GPT-6 Astra solved 98% of complex physics problems and found errors in scientists' answers Researchers tested AI models on 50 unpublished theoretical physics problems, each of which could take an expert almost a whole day to solve. GPT-6 Astra coped with 49 of...
Telegram — GPT Time | Нейросети | ИИ
RUРазработчик протестировал Astra и DeepSeek V4.1-1017 Flash, используя один и тот же промпт и максимальный уровень рас...
Разработчик протестировал Astra и DeepSeek V4.1-1017 Flash, используя один и тот же промпт и максимальный уровень рассуждений. Результаты меня действительно удивили. - Astra справилась за 20 минут, стоимость составила $7. - DeepSeek потребовались 1 час 40 минут и $0,90. Astra…
Telegram — Вайб-кодинг
RUDeepSeek обновила V4.1 Flash
⚡️ DeepSeek обновила V4.1 Flash Версия V4.1-Flash-1017 стала лучше в кодинге и агентских задачах, а также генерирует меньше токенов. На DeepSWE v1.1 модель набрала 77 баллов, обойдя Claude Opus 5 и GPT-5.6 Sol на этом тесте. Параллельно DeepSeek открыла Search API и Fetch API…
Telegram — GPT Time | Нейросети | ИИ
News: AI clones itself, Gemini 4 Argon FOR EVERYONE, Anthropic built its own Git
Neural Network News – Anthropic explores AI worm, while another team connects with a fleet of AI agents. We break down these experiments, new Google Gemini 4 Argon models, Claude Code update, and the story of a voice recorder in the OpenAI office. Comparing Muse Wonder and Muse Wonder Full with…
YouTube — ProdAdvice
OpenCode hack is INSANE.
So, Open Code just got hacked. This is one of the most popular AI coding tools out there. Basically, an open-source version of Cloud Code, and it just got one of the biggest breaches um of this year. And a lot of you are probably impacted. So, here is what the official Open Code…
YouTube — David Ondrej (AI Agents)
Paragon is in touch - summing up the day!
Paragon is in touch - summing up the day! The main thing for the day🟠10.17 🟠Gemini 4 Argon - Minecraft completed in 27 hours 🟠OpenAI - spent test coins without asking 🟠Reflection Beam - a new open model for the code 🟠Anthropic - agents' hours are six times higher than human ones 🟠Claude -...
Telegram — Paragon | Нейросети и ИИ
Why Britain should resist the temptation of cheap and scary Chinese AI
The article warns of the risks of using cheap Chinese AI technology, which could lead to political censorship and surveillance. It is mentioned that companies such as DeepSeek and Xiaomi are rapidly developing their AI solutions, which threatens Western industries and security. Author…
Digital Donut AI Digest
DeepSeek V4.1 Flash 1017 generates 244 tokens per second, surpassing even GPT-6.1 Sol in Ultrafast mode,...
DeepSeek V4.1 Flash 1017 generates 244 tokens per second, outperforming even GPT-6.1 Sol in Ultrafast mode, available at $500 per month! In addition, DeepSeek has provided developers with access to its own web search and page loading tools via an API.…
Telegram — Вайб-кодинг
Deepseek V4.1 Flash 1017 beats larger rivals in agentic tasks while once again massively cutting per-task costs
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 Decoder
The Chinese released DeepSeek 4.1-1017, which is almost equal to Opus 5.
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…
Telegram — Вайб-кодинг
DeepSeek V4.1-Flash-1017: higher metrics, shorter answers, Search and Fetch API open to everyone
🐋 DeepSeek V4.1-Flash-1017: higher metrics, shorter answers, Search and Fetch API open to everyone DeepSeek has updated the Flash version V4.1. The model is stronger than the previous one, solves problems faster and produces shorter answers, that is, it is cheaper. Available on the website, in the application and via API, weights...
Telegram — AI Machine Learning Big Data