![[6399c79e-e753-4b74-87e0-9d97d103eb4a.mp3]]

网信办：中国AI个人用户注册破31亿，模型数量全球第一

简报：

• 国家网信办透露，我国个人用户注册的大模型应用数量已超过31亿，服务通过网页、移动应用、API接口、本地及云服务等多种方式提供。
• 全球已发布3755个大模型，其中中国企业贡献1509个，占全球总数的40%，位居世界首位。
• 截至今年3月，中国已有346款生成式人工智能服务完成备案；到6月，用户使用AI产品进行问答的比例高达80.9%。
• 预计到2024年，我国人工智能产业规模将突破7000亿元，并连续多年保持20%以上的增长率。

相关链接：

• AI大模型用户注册数量突破31亿，服务覆盖面持续扩大

Cyberspace Administration: China's AI Personal User Registrations Exceed 3.1 Billion, Model Count Ranks First Globally

Brief:

• The Cyberspace Administration of China (CAC) revealed that the number of large model applications registered by individual users in China has exceeded 3.1 billion, with services provided through various channels including web pages, mobile apps, API interfaces, local, and cloud services. 🚀
• Globally, 3,755 large models have been released, with Chinese enterprises contributing 1,509 of them, accounting for 40% of the global total and ranking first worldwide. 🥇
• As of March this year, 346 generative AI services in China have completed filing procedures; by June, the proportion of users utilizing AI products for Q&A reached as high as 80.9%.
• It is projected that by 2024, China's artificial intelligence industry scale will exceed 700 billion yuan, maintaining an annual growth rate of over 20% for consecutive years. 📈

Related Link:

• Number of AI Large Model User Registrations Exceeds 3.1 Billion, Service Coverage Continues to Expand

万兴天幕大模型跻身国内前四，并与华为云成立AI视频实验室

简报：根据大模型评测平台SuperCLUE发布的《7月文生视频大模型全球榜单》，万兴科技旗下的天幕音视频多媒体大模型2.0位列国内第四，同时公司已与华为云举行“AI视频大模型实验室”揭牌仪式，双方将共同探索AI在数字创意场景的应用与创新。

相关链接：

• 万兴科技天幕2.0模型成国内第四，携手华为云共建AI视频实验室

Wondershare SkyPaint Large Model Ranks Among Top Four in China, Establishes AI Video Lab with Huawei Cloud

Briefing: According to the "July Text-to-Video Large Model Global Ranking" released by the large model evaluation platform SuperCLUE, Wondershare Technology's SkyPaint Audio-Video Multimedia Large Model 2.0 has been ranked fourth domestically. 🏆 Concurrently, the company has held an unveiling ceremony for the "AI Video Large Model Lab" with Huawei Cloud, where both parties will jointly explore the application and innovation of AI in digital creative scenarios. 💡✨

Related Links:

• Wondershare SkyPaint 2.0 Model Becomes Fourth in China, Collaborates with Huawei Cloud to Build AI Video Lab

阿里发布AI新工具WebShaper，GAIA测试成绩超越GPT-4o

简报：

• 阿里通义实验室发布了WebAgent系列的开源新工具WebShaper，采用创新的“形式化驱动”范式，旨在提升AI处理复杂任务时的信息检索与推理能力。
• 在权威的GAIA基准测试中，基于WebShaper数据集训练的开源模型取得了60.19的高分，其性能表现超越了业界领先的Claude 3.5 Sonnet和GPT-4o。
• WebShaper的核心创新在于其数据集生成框架，通过系统化的任务形式化方法，确保训练数据的知识结构与推理逻辑在语义上高度一致。
• 作为WebAgent开源项目的一部分，WebShaper连同WebSailor等组件已在GitHub上发布，并获得社区广泛关注。

相关链接：

• 项目地址 (GitHub)
• 数据集 (HuggingFace)
• 数据集 (ModelScope)
• 相关报道 (AIbase)

Alibaba Launches New AI Tool WebShaper, Outperforms GPT-4o on GAIA Benchmark

Briefing:

• Alibaba's Tongyi Lab has launched WebShaper, a new open-source tool within its WebAgent series. It adopts an innovative 'formalization-driven' paradigm, aiming to enhance AI's information retrieval and reasoning capabilities when handling complex tasks. 💡
• In the authoritative GAIA benchmark test, the open-source model trained on the WebShaper dataset achieved a high score of 60.19, with its performance surpassing industry leaders Claude 3.5 Sonnet and GPT-4o. 🚀
• WebShaper's core innovation lies in its dataset generation framework, which employs systematic task formalization methods to ensure high semantic consistency between the knowledge structure and reasoning logic within the training data.
• As part of the WebAgent open-source project, WebShaper, along with components like WebSailor, has been released on GitHub and has garnered significant community attention. ⭐

Related Links:

• Project Address (GitHub)
• Dataset (HuggingFace)
• Dataset (ModelScope)
• Related Report (AIbase)

AI视频新突破：Moonvalley让手绘草图一键生成影片

简报：

• AI公司Moonvalley为其Marey视频模型发布了名为“Sketch-to-Video”的新功能。
• 用户通过提供文本描述和手绘草图，即可生成长达5秒、分辨率达1080p的电影级视频片段。
• 该功能基于Marey模型，该模型仅使用完全授权的高清视频进行训练，以避免版权争议，确保商业安全。
• 用户还可通过“相机控制”和“运动控制”功能，对生成的视频镜头和物体轨迹进行调整。
• 此技术旨在降低影视制作门槛与成本，已被多家大型制片厂和广告公司用于试点项目。

相关链接：

• Moonvalley发布Sketch-to-Video功能：手绘草图秒变电影级视频

AI Video Breakthrough: Moonvalley Transforms Hand-Drawn Sketches into Films with One Click

Briefing:

• AI company Moonvalley has released a new feature called "Sketch-to-Video" for its Marey video model. 🚀
• Users can generate cinematic video clips up to 5 seconds long and in 1080p resolution by providing text descriptions and hand-drawn sketches. ✨
• This feature is based on the Marey model, which is trained exclusively on fully licensed high-definition videos to avoid copyright disputes and ensure commercial safety.
• Users can also adjust generated video shots and object trajectories using "Camera Control" and "Motion Control" features.
• This technology aims to lower the barrier and cost of film and television production, and has already been adopted by several major film studios and advertising companies for pilot projects. 🎬

Related Links:

• Moonvalley Releases Sketch-to-Video Feature: Hand-Drawn Sketches Instantly Become Cinematic Videos