昆仑万维正式发布Skywork Deep Research Agent v2

Core Insights - Kunlun Wanwei Technology Co., Ltd. has launched the SkyWorkAI technology release week from August 11 to August 15, introducing a new model each day, including SkyReels-A3, Matrix-Game2.0, Matrix-3D, and SkyworkUniPic2.0 [1] - The SkyworkDeepResearchAgentv2, released on August 14, serves as the core engine for the Skywork Super Agents, significantly enhancing the role of large models in the AI Office sector by producing high-density information documents, PPTs, and spreadsheets [1][2] - The new version integrates multi-modal retrieval, understanding, and generation capabilities, marking the industry's first "multi-modal deep research" agent [1][2] Technical Breakthroughs - The Skywork team achieved advancements in four key areas: multi-modal crawling technology (MM-Crawler), long-distance multi-modal information collection, asynchronous parallel multi-agent multi-modal understanding architecture, and multi-modal result presentation capabilities [2] - The SkyworkDeepResearchAgentv2 introduces a "multi-modal deep browser agent," transforming social media content analysis and data insights with features like low latency, high response rates, and flexible decision-making [2][3] Performance and Capabilities - The SkyworkBrowserAgent can simulate human browsing and interaction, revolutionizing traditional data collection and analysis methods, effectively addressing multiple pain points of conventional browser agents [3] - The SkyworkDeepResearchAgentv2 has enhanced deep information search and complex task execution capabilities, achieving state-of-the-art (SOTA) results across various task evaluation sets [3] - The agent's accuracy improves with increased thinking time in a parallel thinking mode, showcasing the potential and scalability of the self-developed system architecture [3]