China’s DeepSeek Rolls Out a Long-Anticipated Update of Its AI Model

Source Domain: www.usnews.com

Here are some key points from the article about DeepSeek’s latest AI model release:

DeepSeek V4 Launch: DeepSeek, a Chinese AI startup, has launched preview versions of its new model V4.
Competitiveness: DeepSeek’s V4 is anticipated to compete strongly with AI models from U.S. companies like OpenAI’s ChatGPT, Anthropic’s Claude, and Google’s Gemini.
Features & Improvements: DeepSeek highlighted significant advancements in knowledge, reasoning, and agentic capabilities in V4, alongside an open-source approach.
Chip Independence: The new models are partially supported by Huawei chips, reducing dependence on U.S. suppliers like Nvidia.
Benchmark Performance: DeepSeek claims V4 Pro Max outperforms other U.S. models on reasoning benchmarks, although it doesn’t lead in all measures.
Industry Viewpoint: Analysts like Lian Jye Su from Omdia believe V4 positions DeepSeek competitively against U.S. rivals, while some skeptics like Ivan Su from Morningstar believe significant breakthroughs comparable to its predecessor, R1, are yet to be seen.
Allegations from Competitors: Anthropic and OpenAI have accused DeepSeek of illicitly attempting to improve its models by mimicking their technology, a claim rejected and refuted by China’s embassy in Washington.
Global Influence: DeepSeek’s increasing use, particularly where Huawei phones are prevalent, has gained traction in several developing nations.