目录
![[7571893c-7e42-499a-bb77-a1fb5fe4bbd0.mp3]]
百度AI搜索月活达3.22亿，领跑国内市场
简报：
- QuestMobile《2025年AI应用市场半年报》显示，百度AI搜索月活跃用户规模高达3.22亿，位居国内AI搜索行业第一。
- 8月6日，百度搜索对智能框进行升级，支持更长、更复杂的内容输入，并能便捷调用AI视频、AI解题、AI写作、AI PPT、AI画图等多种AI工具。
- 目前该升级已在百度App端上线，PC端将于8月11日完成全量更新。
相关链接：
Baidu AI Search MAU Reaches 322 Million, Leading Domestic Market
Brief:
- QuestMobile's "2025 H1 AI Application Market Report" shows that Baidu AI Search's monthly active user base has reached an impressive 322 million, ranking first in China's AI search industry. 🚀
- On August 6, Baidu Search upgraded its smart input box to support longer, more complex content input and enable convenient access to various AI tools like AI video, AI problem-solving, AI writing, AI PPT, and AI image generation. ✨
- The upgrade is currently live on the Baidu App, with the PC version scheduled for a full update on August 11. 💻
Related Links:
Anthropic发布Claude Opus 4.1，编码能力创新高
简报：
- Anthropic于2025年8月5日正式发布其旗舰AI模型Claude Opus 4.1，作为Opus 4的升级版。
- 新模型在软件工程基准测试（SWE-bench）上的准确率达到74.5%，刷新了行业记录，同时在代理性能和安全性方面也有所提升。
- Claude Opus 4.1现已通过API向付费用户和开发者提供，定价与其前代产品保持一致。
相关链接：
Anthropic Releases Claude Opus 4.1, Setting New Heights in Coding Capability 🚀
Briefing:
- On August 5, 2025, Anthropic officially launched its flagship AI model, Claude Opus 4.1, as an upgraded version of Opus 4.
- The new model achieved an accuracy rate of 74.5% in the Software Engineering Benchmark (SWE-bench) test, setting a new industry record, while also improving agent performance and safety. ✨
- Claude Opus 4.1 is now available via API to paid users and developers, with pricing consistent with its predecessor. 🤖
Related Links:
ChatGPT用户增长迅猛，周活将达7亿
简报：
- OpenAI于2025年8月4日表示，ChatGPT的周活跃用户数预计将于本周达到7亿，相较去年同期增长4倍。
- 该数据相比2025年3月底增加了2亿活跃用户，统计范围涵盖免费版、Plus、企业版等所有ChatGPT产品。
- 截至今年7月，全球用户每日与ChatGPT的互动信息超过25亿条。
- 企业与教育机构的付费用户数在8月初已达到500万，而2024年末时该数据约为100万。
相关链接：
ChatGPT User Growth Surges, Weekly Active Users Expected to Reach 700 Million
Brief:
- OpenAI stated on August 4, 2025, that ChatGPT's weekly active users are projected to reach 700 million this week, marking a four-fold increase compared to the same period last year. 🚀
- This figure represents an increase of 200 million active users since the end of March 2025, with the count encompassing all ChatGPT products, including the free version, Plus, and Enterprise editions.
- As of July this year, global users interact with ChatGPT more than 2.5 billion times daily. 💬
- The number of paid users from businesses and educational institutions reached 5 million in early August, a significant jump from approximately 1 million at the end of 2024. 📈
Related Links:
xAI下一代模型Grok-2即将开源
简报：
- 埃隆·马斯克宣布，其人工智能公司xAI开发的下一代大语言模型Grok-2，计划于下周进行开源。
相关链接：
xAI's Next-Gen Model Grok-2 Soon to Be Open-Sourced 🚀
Brief:
- Elon Musk announced that Grok-2, the next-generation large language model developed by his AI company xAI, is slated for open-sourcing next week. ✨
Related Link:
GPT-5震撼亮相：支持文本生成视频，逻辑推理能力再突破
简报：
- OpenAI CEO山姆·奥特曼（Sam Altman）首次公开展示了GPT-5模型的内部实测情况。
- 据演示，GPT-5具备强大的文本到视频生成能力，能够根据文本提示直接生成长达1分钟的高质量视频。
- 该模型在逻辑推理方面也实现了显著突破，展现了更强的复杂问题解决能力。
相关链接：
GPT-5 Makes a Stunning Debut: Text-to-Video Generation Supported, Logical Reasoning Reaches New Heights
Brief:
- OpenAI CEO Sam Altman has for the first time publicly showcased internal test results of the GPT-5 model. 🚀
- The demonstration revealed GPT-5's powerful text-to-video generation capability, enabling it to directly create high-quality videos up to 1 minute in length from text prompts. 🎬
- The model has also achieved a significant breakthrough in logical reasoning, showcasing enhanced capabilities in solving complex problems. 🧠
Related Links:
AMD与高通宣布支持OpenAI新开放模型，赋能端侧AI
简报：
- OpenAI于8月5日以Apache 2.0许可证形式，发布了两款gpt-oss系列开放推理模型：gpt-oss-20b和gpt-oss-120b。
- AMD宣布旗下锐龙 AI Max+ 395处理器支持gpt-oss-120b模型，Radeon RX 9070 16GB显卡支持gpt-oss-20b模型。
- AMD宣称锐龙 AI Max+ 395是全球首款能运行gpt-oss-120b模型的消费级AI PC处理器，输出速率可达30 Token/s。
- 高通宣布其旗舰骁龙平台支持gpt-oss-20b模型，并表示早期测试显示该模型在端侧设备上具有出色的思维链推理表现。
相关链接：
AMD and Qualcomm Announce Support for New OpenAI Open Models, Powering Edge AI 🚀
Briefing:
- OpenAI released two gpt-oss series open inference models, gpt-oss-20b and gpt-oss-120b, under the Apache 2.0 license on August 5th.
- AMD announced that its Ryzen AI Max+ 395 processor supports the gpt-oss-120b model, and its Radeon RX 9070 16GB graphics card supports the gpt-oss-20b model.
- AMD claims the Ryzen AI Max+ 395 is the world's first consumer-grade AI PC processor capable of running the gpt-oss-120b model, achieving an output rate of up to 30 Tokens/s. 🌟
- Qualcomm announced that its flagship Snapdragon platform supports the gpt-oss-20b model, stating that early tests show excellent chain-of-thought reasoning performance on edge devices. 🔥
Related Links:
谷歌DeepMind发布Genie 2，可从图文生成3D互动世界
简报：
- 谷歌DeepMind团队发布了名为Genie 2的基础世界模型，是其先前模型Genie的继任者。
- 该模型能够基于单张图像或文本描述，生成可供人类或AI智能体进行实时互动的3D环境。
- 在演示中，Genie 2与Imagen 3模型结合，先根据文本生成初始图片，再将图片转化为可交互的3D场景。
- 根据DeepMind的说法，Genie 2目前能生成长达60秒的“一致性”世界。
相关链接：
- https://www.aibase.com/zh/news/20270
- https://www.aibase.com/zh/news/20262
- https://www.aibase.com/zh/news/20266
Google DeepMind Releases Genie 2, Capable of Generating Interactive 3D Worlds from Images and Text 🌐✨
Brief:
- The Google DeepMind team has released Genie 2, a foundational world model and the successor to its previous model, Genie.
- This model can generate 3D environments, interactable in real-time by humans or AI agents, based on a single image or text description.
- In demonstrations, Genie 2 is combined with the Imagen 3 model, first generating an initial image from text, then converting that image into an interactive 3D scene.
- According to DeepMind, Genie 2 can currently generate "consistent" worlds lasting up to 60 seconds. ⏳
Related Links:
- https://www.aibase.com/zh/news/20270
- https://www.aibase.com/zh/news/20262
- https://www.aibase.com/zh/news/20266
如果对你有用的话，可以打赏哦
打赏
本文作者:topwind
本文链接:
版权声明:本博客所有文章除特别声明外，均采用 BY-NC-SA 许可协议。转载请注明出处！
目录