Ein KI-Rechenzentrum entsteht aktuell in Lübbenau, Brandenburg - für 11 Milliarden Euro (Symbolbild). – Alle Rechte vorbehalten: IMAGO / Jochen Eckel; Bearbeitung: netzpolitik.orgEin neuer UN-Report warnt vor dem wachsenden Energie- und Wasserverbrauch von Rechenzentren aufgrund des KI-Booms. Doch statt auf die Verantwortung von Tech-Konzernen zu pochen, gibt er Tipps für Nutzer:innen, wie sie…
Empresa lançou o Serviço de Model as a Service (MaaS) com foco em baixa latência, alta confiabilidade e aplicações corporativas em escala.
Inteligência artificial generativa faz parte do cotidiano de praticamente qualquer usuário da internet atualmente – para uma simples busca no Google, há uma IA envolvida. Mesmo que você não acesse um chatbot convencional, é provável que já tenha interagido com alguma IA de outra maneira.
Contudo, existe uma certa confusão sobre a necessidade de assinar uma IA generativa, principalmente acerca…
I am a teacher. Since January 2013, I have been teaching the “Operating Systems” course at the Engineering Faculty of
UNAM. And yes, that means May and November are highly stressful months, where I have to review the work done by my students and… _sigh_ … come to the difficult decisions leading to a numerical score that will, in very very short, represent the 64 hours they spent listening to me…
O chatbot de inteligência artificial (IA) DeepSeek acirrou a concorrência no mercado em termos de preço. Nos últimos dias, a plataforma anunciou um corte agressivo de preços para a utilização de tokens de um dos seus mais recentes modelos de linguagem.
A medida afeta o V4‑Pro, que estava em fase promocional com 75% de desconto pelo uso. A partir de 31 de maio de 2026, quando ele voltaria ao…
中國AI新創DeepSeek傳洽談百億美元融資，估值上看450億美元。創辦人將個人大額出資，並堅持開源與AGI路線，拒絕短期變現，此舉旨在囤積算力與人才，挑戰全球AI產業格局。
就在全球AI業者為大型語言模型軍備競賽瘋狂加碼之際，以開源策略與極高性價比震撼業界的中國AI新星DeepSee […]
The potential investment by CATL comes as the company expands into new businesses.
For details, please visit CnEVPost.
While proprietary AI models such as OpenAI’s ChatGPT and Google Gemini remain popular, the tide may be shifting to open models as IT leaders move to customize AI and control costs.
Sometimes known as “open-weight models,” the alternatives to large language models (LLMs) can provide decision-makers with better visibility and control over internal AI use, which closed models do not, analysts said.…
Elon Musk and Sam Altman may grab the headlines, but billionaires, theorists, and futurists are designing the most transformative technology of our era
A Hangzhou judges declares illegal the termination of the contract of an employee whose job was replaced with large language models
AI知識蒸餾已成美中科技競爭新焦點。此技術涉及模型能力轉移，引發智慧財產權與國安疑慮，使競爭從硬體擴展至演算法。其技術定性、影響評估與監管框架，已成關鍵地緣政治挑戰。
中國 AI 應用程式「豆包」因應高昂算力成本，宣布推出付費訂閱服務。其最高階專業版年費達人民幣5088元，引發熱議，但仍會保留免費版本供使用者繼續使用。
Elon Musk y Sam Altman copan los titulares, pero millonarios, teóricos y futurólogos diseñan la tecnología más transformadora de nuestra era
人工智慧崛起引發中國家長教育焦慮，憂心當前應試體系將如晚清廢科舉般瓦解。此變革促使社會反思傳統教育價值，並為下一代探索適應人機協作的未來。
中國新創 DeepSeek 釋出 DeepSeek-V4 預覽版，針對華為昇騰晶片優化。該模型支援百萬級 Token 脈絡長度與強化代理人能力，展現中國 AI 產業鏈自主化，旨在減少對 NVIDIA 晶片依賴。
白宮將與美企合作，反制中國以「蒸餾」技術竊取AI成果。雙方將分享情資、協同抵禦，並探索究責方式，以嚇阻此類工業規模的惡意活動。
Novo modelo mantém estrutura de código aberto e usa chips da Huawei para driblar o domínio da americana Nvidia.
DeepSeek V4: startup chinesa revela nova IA um ano após surpreender o mercado
A China rejeitou as acusações feitas pelo governo estadunidense de que entidades chinesas estariam extraindo ilegalmente capacidades de modelos de inteligência artificial desenvolvidos nos Estados Unidos. O porta-voz do Ministério das Relações Exteriores, Guo Jiakun, classificou as alegações como “completamente infundadas” e afirmou que são “difamação contra as conquistas da China em sua…
DeepSeek a profité du lancement de GPT-5.5 par OpenAI pour sortir V4, un modèle open source de 1,6 billion de paramètres qui coûte 7 à 9 fois moins cher que ses rivaux américains. Le genre de timing qu'on ne choisit pas par hasard.
Çin merkezli yapay zeka şirketi DeepSeek, bugün “V4” serisi yeni açık yapılı modellerini duyurdu. Yeni ailede V4-Pro ve V4-Flash olmak üzere iki ana model yer alıyor. 1,6 trilyon toplam ve 49 milyar aktif parametreye sahip olan V4-Pro, şirketin açıklamasına göre açık formda dünyanın en iyi kapalı modelleriyle yarışabilecek düzeyde performans ortaya koyuyor. Daha hızlı ve ekonomik bir seçenek…
DeepSeek ci riprova. Un anno dopo l’uscita del suo chatbot che ha rivoluzionato il settore, la startup cinese lancia un nuovo modello a basso costo ma in grado di competere con i rivali statunitensi. DeepSeek-V4 “presenta un contesto estremamente lungo di un milione di parole, raggiungendo la leadership sia nel settore nazionale che in quello […]
Poco más de 12 horas después de que OpenAI presentara ChatGPT 5.5, su competencia directa ha respondido anunciando un nuevo modelo que espera rivalizar con la firma de Sam Altman. Se trata de DeepSeek V4-Pro y V4-Flash. Te contamos todos los detalles.
Desde hace algunos meses, nos hemos acostumbrado a las constantes presentaciones de las principales empresas de Inteligencia Artificial que…
DeepSeek unveils V4 Pro and Flash models with 1M context, advanced reasoning, agent integration, and improved efficiency. New architecture targets scalable AI performance and API migration by 2026.
The post DeepSeek Unveils V4 Model Series: High-Parameter AI Push Targets Efficiency And Frontier Performance appeared first on Metaverse Post.
DeepSeek-V4 ออกแล้ว เทียบชั้น Opus 4.6, Gemini 3.1 Pro
Body
DeepSeek ออกโมเดลเวอร์ชัน 4 สองตัว ได้แก่ DeepSeek-V4-Pro และ DeepSeek-V4-Flash ชูจุดเด่นว่ามีความเก่งในหมวดการเขียนโปรแกรมระดับเดียวกับโมเดลชั้นนำ เช่น Opus 4.6, GPT-5.4 xhigh, และ Gemini 3.1 Pro (high)
DeepSeek-V4-Pro เป็นโมเดลขนาด 1.6T-A49B แบบ FP4+FP8 ความสามารถใกล้เคียงกับโมเดลชั้นนำแต่ก็ยังห่างจากโมเดลใหม่มากๆ เช่น Claude Opus…
The cyber capabilities of AI models have experts rattled. AI’s social skills may be just as dangerous.
美中 AI 競賽呈現「大腦」與「身體」之爭。美國以晶片及模型領先，中國則在機器人領域佔優。雙方正加速整合軟硬體，競逐技術主導權，勝負未定。
GitHub's March 2026 updates introduce secret scanning for AI agents via MCP, 37 new detectors, and expanded push protection. Learn how to secure AI-generated code.
EL PAÍS reconstructs what many media outlets sold as ‘the news story of the year,’ and some of its protagonists described as a ‘disgrace’: the supposed ‘gold medal-level performance’ by big tech, which overshadowed the real competition involving 600 youngsters
買収されても、国は出られない。中国発で、現在はシンガポールを拠点とするAIエージェント開発企業Manusが、中国当局の審査に直面しています。Financial Timesが報じ、ReutersやWashington Postなどが後追いしました。中国当局は、ManusのCEO X...
smhn.infoにアクセスすると、全文を読むことができます。