DeepSeek新模型突破深度推理极限，人工智能AIETF（515070）持仓股中科星图大涨超4%

Core Insights - The A-share market is currently in a consolidation phase, with slight increases in sectors such as artificial intelligence, AI glasses, and CPO, while humanoid robots and storage chips are underperforming [1] - DeepSeek has launched a new mathematical reasoning model, DeepSeekMath-V2, which demonstrates exceptional performance in international mathematics competitions and utilizes an innovative self-verification framework [1][2] - The domestic AI computing power is rising, with significant advancements in large models and applications, reshaping the global competitive landscape [2] Group 1: Market Performance - The A-share market indices are consolidating, with specific sectors like artificial intelligence and AI glasses showing slight gains [1] - The AI ETF (515070) experienced a minor decline of 0.28%, with leading stocks including 360, Zhongke Xingtu, and Hengxuan Technology [1] Group 2: Technological Developments - DeepSeekMath-V2, a new model by DeepSeek, features a "generator-verifier" architecture for rigorous self-examination of mathematical proofs, ensuring the integrity of reasoning processes [1] - The model is built on DeepSeek-V3.2-Exp-Base and is open-sourced on platforms like Hugging Face and GitHub under the Apache 2.0 license [1] Group 3: Industry Trends - Domestic AI is making significant strides in computing power, large models, and applications, reducing reliance on foreign manufacturers and reshaping the competitive landscape [2] - Major domestic AI models such as Deepseek, Tongyi Qianwen, Byte Bean, and Kimi are rapidly improving and becoming leaders in the global AI model competition [2] - The AI ETF (515070) tracks the CS AI theme index and includes key stocks that provide technology, foundational resources, and applications in the AI industry [2]