DEEPSEEK

About DeepSeek

Chinese AI research lab building competitive open-source LLMs. Known for DeepSeek Coder and strong performance on coding benchmarks.

Recent News about DeepSeek

• China AI core industry tops ¥1T as MIIT sets 2026 priorities

  On December 26, 2025, China’s Ministry of Industry and Information Technology reported at its national work conference that the country’s AI core industry has surpassed 1 trillion yuan in size. Officials also highlighted tens of thousands of high‑quality datasets, large domestic compute clusters, and competitive local foundation models such as DeepSeek and Tongyi Qianwen.

  Source: China News Service (Chinanews)2025-12-26T12:23:58.000Z

• China records 700+ filed generative AI models under ‘AI+’ push

  On December 26, 2025, People’s Daily reported that over 700 generative AI large-model products have completed official filing in China during the 14th Five-Year Plan period. Officials also highlighted major progress in integrated circuits, AI, basic software and a rising share of the digital economy in GDP.

  Source: People’s Daily Online2025-12-25T22:20:00.000Z

• Investors pivot to Chinese AI stocks amid US bubble fears

  On December 23, 2025, Reuters reported that global asset managers are increasing allocations to Chinese AI companies and ETFs as concerns mount over lofty U.S. AI valuations. Funds are rotating into Chinese tech names like Alibaba, Baidu, Tencent and domestic AI chipmakers such as Cambricon, Moore Threads and MetaX, helped by Beijing’s push for AI self‑reliance.

  Source: Reuters2025-12-23T07:28:00.000Z

• DeepSeek and Xiaomi face US push onto Pentagon watchlist

  On December 22, 2025, nine U.S. Republican lawmakers sent a letter urging the Pentagon to add Chinese AI startup DeepSeek, smartphone maker Xiaomi and several other tech firms to a Defense Department list of entities allegedly aiding China’s military. The request targets inclusion on the Section 1260H list, which flags companies seen as linked to China’s armed forces but does not itself impose sanctions.

  Source: Reuters2025-12-22T20:16:00.000Z

• New York RAISE Act sets tough transparency rules for frontier AI

  New York Governor Kathy Hochul has signed the RAISE Act into law, requiring large frontier AI developers to publish safety frameworks and report serious incidents within 72 hours. The law, announced December 19, 2025 and now being widely reported on December 20–21, 2025, also creates a new AI oversight office in the state’s Department of Financial Services and authorizes multimillion‑dollar penalties for non‑compliance.([governor.ny.gov](https://www.governor.ny.gov/news/governor-hochul-signs-nation-leading-legislation-require-ai-frameworks-ai-frontier-models?utm_source=openai))

  Source: Engadget2025-12-21T02:04:51.690Z

• New York RAISE Act sets frontier AI safety rules for big models

  On December 20, 2025, New York Governor Kathy Hochul signed the RAISE Act, creating one of the strictest AI safety laws in the United States. The law forces large frontier AI developers to publish safety frameworks, assess critical risks and report serious incidents to the state within 72 hours.

  Source: TechCrunch2025-12-20T18:20:00.000Z

• China’s AI music tools reshape songwriting and discovery

  A People’s Daily feature on Dec. 19 describes how China’s digital music base in Hangzhou and companies like Tencent Music are using large models such as DeepSeek and tools like Suno V4 to automate songwriting and personalize music recommendations. The article cites examples where full songs are generated in minutes and notes Tencent’s AI songwriting feature has produced tens of millions of tracks with over a billion plays.([paper.people.com.cn](https://paper.people.com.cn/rmrbhwb/pc/content/202512/19/content_30125699.html))

  Source: People’s Daily Overseas Edition2025-12-19T22:05:21.022Z

• CE.cn says Chinese open‑source LLMs are reshaping global AI access

  China Economic Net published an analysis on December 18, 2025 arguing that Chinese open‑source models like DeepSeek and Alibaba’s Qwen now underpin nearly 30% of global AI usage. Citing a16z and Stanford data, the piece says Chinese models are gaining share on Hugging Face and powering deployments across the Global South due to lower costs and on‑premise sovereignty. ([en.ce.cn](https://en.ce.cn/Insight/202512/t20251218_2650721.shtml?utm_source=openai))

  Source: China Economic Net2025-12-18T03:28:00.000Z

• Study finds Korean LLMs trail global rivals even on home‑turf exam math

  Korea JoongAng Daily reports that a Sogang University research team tested five Korean large language models against leading US and Chinese models on College Scholastic Ability Test (CSAT) questions and advanced essay‑style math problems, finding a stark performance gap. Foreign models like GPT‑5.1, Gemini 3 Pro Preview, Claude Opus 4.5, Grok 4.1 Fast and DeepSeek V3.2 scored between 76 and 92 points, while domestic systems such as Upstage’s Solar Pro‑2, LG AI Research’s Exaone 4.0.1, Naver’s HCX‑007, SK Telecom’s A.X 4.0 and NCSoft’s Llama Varco 8B mostly landed in the 20‑point range, with the weakest model scoring just 2 points. Even when models were allowed to use Python tools and given multiple attempts on a separate EntropyMath benchmark, non‑Korean systems still dominated, suggesting that Korea’s sovereign‑AI push has not yet closed the reasoning gap with frontier models. The authors say they plan to build an international math leaderboard and update tests as newer model versions arrive, but the results reinforce concerns that language‑localized LLMs can lag badly on deeper problem‑solving unless they match global players in both data and algorithmic sophistication.

  Source: Korea JoongAng Daily2025-12-15T05:11:00.000Z

• Mistral launches open‑weight 'frontier AI family' to challenge DeepSeek and US rivals

  French startup Mistral released a four‑model "frontier AI family," including a 675‑billion‑parameter sparse Mixture‑of‑Experts model dubbed Mistral Large 3, all under the permissive Apache 2.0 open‑weight license. The models, trained on thousands of Nvidia H200 GPUs, are pitched as state‑of‑the‑art multimodal and multilingual systems that enterprises can run and fine‑tune locally, giving privacy‑sensitive European customers an alternative to Chinese lab DeepSeek and US closed‑weight offerings from OpenAI and Google.

  Source: Decrypt2025-12-03T22:07:08.704Z