新加坡国家AI计划放弃Meta模型,转向阿里千问;腾讯混元OCR模型宣布开源丨AIGC日报
创业邦·2025-11-26 00:08

Group 1 - Singapore's National AI Strategy is shifting from Meta's model to Alibaba's Qwen, indicating a significant expansion of Chinese open-source AI models globally. Since its full release in 2023, Qwen has surpassed models like Llama and Deepseek, becoming the strongest and most widely used open-source large model, with over 600 million downloads globally [2] - The U.S. plans to establish a comprehensive AI platform, likened to the Manhattan Project, to accelerate AI applications in transformative scientific discoveries. This initiative, called the "Genesis Mission," aims to integrate resources from national laboratories, scientists, businesses, and universities to train foundational models using federal scientific datasets [2] - Elon Musk is advancing the use of Grok to replace employees at X company, significantly reducing the team responsible for trust and safety issues. The team has shrunk from over 100 members to fewer than 20, as Musk plans to fully automate user interest matching through Grok [2] - Tencent has launched the HunyuanOCR model, which is open-source and built on its native multi-modal architecture, achieving state-of-the-art results in various OCR applications with only 1 billion parameters [2]