DeepSeek

Twitter

Facebook

Copy Link

Key Facts

OrganizationDeepSeek

Latest ReleaseV4 Pro, V4 Flash (2026)

TypeMultimodal LLM

Initial Release Year2023

Top News

Samsung accidentally reveals AI-powered Bixby overhaul, then deletes the entire announcement

Samsung unveiled a major Bixby reboot powered by Perplexity AI but mysteriously pulled the announcement hours later. The update promises natural language understanding and real-time web search capabilities, set to debut with One UI 8.5 and the Galaxy S26 series. The leaked details reveal Samsung's strategy to revive its struggling voice assistant.

Technology10 sources

Samsung accidentally reveals AI-powered Bixby overhaul, then deletes the entire announcement

Technology10 sources

Chinese AI startups surge with investor backing despite chip controls and profitability hurdles

China's AI boom accelerates as startups raise hundreds of millions in funding rounds, with Zhipu AI and MiniMax making stellar Hong Kong stock exchange debuts. But beneath the excitement, Chinese AI faces critical challenges: US export controls on microchips, mounting losses, and the puzzle of achieving profitability while competing with OpenAI and other Western rivals.

Business4 sources

Chinese AI startups surge with investor backing despite chip controls and profitability hurdles

Business4 sources

DeepSeek's Engram tackles AI compute waste with conditional memory breakthrough

DeepSeek unveiled Engram, a conditional memory technique that separates static information retrieval from complex reasoning tasks. The method allows AI models to bypass GPU memory constraints by committing knowledge to system RAM, reducing reliance on expensive high-bandwidth memory while improving performance on long-context queries.

Technology5 sources

DeepSeek's Engram tackles AI compute waste with conditional memory breakthrough

Technology5 sources

DeepSeek V4 targets mid-February launch with coding capabilities to challenge OpenAI and Anthropic

Chinese AI startup DeepSeek is reportedly preparing to launch its V4 model around mid-February, with internal tests suggesting it could outperform both Anthropic's Claude and OpenAI's GPT series in coding tasks. The model features breakthroughs in handling extremely long code prompts and introduces the innovative Ingram architecture, which separates memory from reasoning to reduce GPU dependency and improve efficiency.

Technology5 sources

DeepSeek V4 targets mid-February launch with coding capabilities to challenge OpenAI and Anthropic

Technology5 sources

DeepSeek AI gains traction in developing nations as free models fuel adoption gap concerns

Chinese AI startup DeepSeek is rapidly expanding across the Global South, achieving 89% market share in China and significant adoption in Russia, Belarus, and Africa. A new Microsoft report reveals how the company's free and open-source models are reshaping AI adoption patterns while raising questions about geopolitical influence and the widening technology divide between advanced and developing economies.

Business6 sources

DeepSeek AI gains traction in developing nations as free models fuel adoption gap concerns

Business6 sources

Two Chrome Extensions Caught Stealing AI Chats from 900,000 ChatGPT and DeepSeek Users

Cybersecurity researchers uncovered two malicious Chrome extensions that exfiltrated ChatGPT and DeepSeek conversations from over 900,000 users. The extensions disguised themselves as legitimate AI tools while secretly harvesting AI chatbot conversations and browsing data every 30 minutes. This attack method, dubbed Prompt Poaching, raises concerns about user privacy and corporate espionage as even featured extensions on official stores prove vulnerable.

Technology3 sources

Two Chrome Extensions Caught Stealing AI Chats from 900,000 ChatGPT and DeepSeek Users

Technology3 sources

DeepSeek unveils mHC architecture that could reshape how developers train advanced AI models

Chinese AI firm DeepSeek has released research on Manifold-Constrained Hyper-Connections (mHC), a new AI training method that addresses signal degradation in neural networks. The framework could enable smaller developers to build frontier models without massive computational costs, potentially previewing the architecture behind the anticipated DeepSeek R2 model expected around February's Spring Festival.

Science7 sources

DeepSeek unveils mHC architecture that could reshape how developers train advanced AI models

Science7 sources

AI Boom Creates Over 50 New Billionaires as Startup Investment Surges Past $200 Billion

The artificial intelligence sector generated unprecedented wealth in 2025, minting more than 50 new billionaires as investors poured $202 billion into AI companies. From data labeling startups to AI agents replacing workers, founders saw their net worth surge to ten-figure amounts in record time, with some reaching billionaire status in less than three years.

Business6 sources

AI Boom Creates Over 50 New Billionaires as Startup Investment Surges Past $200 Billion

Business6 sources

Sam Altman warns Google remains huge threat as OpenAI braces for Code Red alerts twice yearly

OpenAI CEO Sam Altman revealed the company will enter Code Red emergency mode once or twice a year as AI competition intensifies. Speaking on the Big Technology Podcast, Altman called Google a huge threat despite OpenAI's lead, noting the tech giant could have 'smashed' them in 2023. The admission comes as Google's Gemini 3 and China's DeepSeek trigger strategic shifts at OpenAI.

Technology3 sources

Sam Altman warns Google remains huge threat as OpenAI braces for Code Red alerts twice yearly

Technology3 sources

NVIDIA AI Servers Achieve 10x Performance Boost for Mixture of Experts Models from DeepSeek and Moonshot

NVIDIA has shattered performance records with its GB200 Blackwell NVL72 servers, delivering a tenfold efficiency boost for mixture-of-expert AI models including those from China's DeepSeek and Moonshot AI. The breakthrough addresses a critical computing bottleneck in scaling MoE models, using 72 chips with high-bandwidth links and 30TB of fast shared memory to enable expert parallelism at unprecedented levels.

Technology3 sources

NVIDIA AI Servers Achieve 10x Performance Boost for Mixture of Experts Models from DeepSeek and Moonshot

Technology3 sources

DeepSeek releases open-source AI models that rival GPT-5 and Gemini at fraction of the cost

Chinese AI company DeepSeek launched two new reasoning models, V3.2 and V3.2-Speciale, claiming performance on par with OpenAI's GPT-5 and Google's Gemini 3.0 Pro. The open-source models cost $0.028 per million tokens compared to Gemini 3's $4.00, and achieved gold-level performance in elite international math and coding competitions. The release challenges American tech dominance while making frontier AI capabilities freely accessible.

Technology13 sources

DeepSeek releases open-source AI models that rival GPT-5 and Gemini at fraction of the cost

Technology13 sources

$DeepSeek releases open-source AI models that rival GPT-5 and Gemini at fraction of the cost$

Tencent Reports Strong Q3 Growth Driven by AI Integration Across Gaming and Advertising

Chinese tech giant Tencent posted 15% quarterly revenue growth, beating analyst expectations as AI enhancements boosted gaming performance and advertising targeting capabilities. The company continues expanding AI investments amid US-China tech tensions.

Business2 sources

Tencent Reports Strong Q3 Growth Driven by AI Integration Across Gaming and Advertising

Business2 sources

China's Military AI Advancements: DeepSeek-Powered Autonomous Vehicles and Nvidia Chip Reliance

China unveils an AI-powered autonomous military vehicle, highlighting its progress in AI for defense. Despite export restrictions, Chinese military institutions continue to use Nvidia chips, raising concerns about technological dependencies and the global AI arms race.

Technology3 sources

China's Military AI Advancements: DeepSeek-Powered Autonomous Vehicles and Nvidia Chip Reliance

Technology3 sources

DeepSeek-OCR: Revolutionary AI Model Compresses Text into Images, Transforming Language Processing

DeepSeek's new open-source AI model, DeepSeek-OCR, introduces a groundbreaking approach to text processing by converting text into images. This method achieves up to 20x compression while maintaining high accuracy, potentially revolutionizing AI language models' efficiency and capabilities.

Technology5 sources

DeepSeek-OCR: Revolutionary AI Model Compresses Text into Images, Transforming Language Processing

Technology5 sources

AI Models Battle in Crypto Trading Showdown: DeepSeek and Grok Lead the Pack

A real-money AI crypto trading competition pits top language models against each other, with DeepSeek and Grok emerging as early leaders. The contest highlights the potential and risks of AI in financial markets.

Technology2 sources

AI Models Battle in Crypto Trading Showdown: DeepSeek and Grok Lead the Pack

Technology2 sources

Prev 1 23Next