![[71d59eaf-6168-4669-b111-883d11d1c1ac.mp3]]

OpenAI新一代GPT-5模型预计将于8月初发布，推出多种版本并整合高级推理能力

简报：

• OpenAI预计将在2025年8月初正式发布下一代语言模型GPT-5，计划同时推出mini和nano等多个版本，并通过API向开发者开放。
• OpenAI CEO山姆・阿尔特曼在播客及社交媒体中提到，GPT-5发布进展顺利，并透露其具备整合o3推理能力、融合多项现有技术，显著提升用户体验。
• 此前GPT-5曾计划于5月推出，因额外测试和延迟顺延至8月，OpenAI已为发布做好准备，但目前尚未发布官方声明。
• 除GPT-5外，OpenAI计划7月底前发布具备先进推理能力的开放权重模型，为开发者和用户带来更多可选AI工具。

相关链接：

• OpenAI prepares to launch GPT-5 in August（The Verge）
• OpenAI's GPT-5 is coming in early August, per sources（MSN）
• OpenAI plans to combine multiple models into GPT-5（BleepingComputer）
• AIBase新闻：OpenAI 即将发布 GPT-5，预计在八月亮相
• IT之家：OpenAI 预计 8 月初正式发布新一代 GPT-5 大模型

OpenAI's Next-Generation GPT-5 Model Expected to Launch in Early August 2025, Featuring Multiple Versions and Advanced Reasoning Capabilities

Briefing:

• OpenAI is expected to officially release its next-generation language model, GPT-5, in early August 2025. 🚀 It plans to launch multiple versions, including 'mini' and 'nano', and make them available to developers via API.
• OpenAI CEO Sam Altman has mentioned in podcasts and on social media that the GPT-5 release is progressing smoothly. He revealed that it will integrate 'o3' reasoning capabilities and combine multiple existing technologies, significantly enhancing user experience. ✨
• Previously, GPT-5 was planned for a May launch but was postponed to August due to additional testing and delays. OpenAI is prepared for the release, though an official statement has not yet been issued.
• In addition to GPT-5, OpenAI plans to release an open-weight model with advanced reasoning capabilities by the end of July, offering developers and users more optional AI tools. 🛠️

Related Links:

• OpenAI prepares to launch GPT-5 in August (The Verge)
• OpenAI's GPT-5 is coming in early August, per sources (MSN)
• OpenAI plans to combine multiple models into GPT-5 (BleepingComputer)
• AIBase新闻：OpenAI 即将发布 GPT-5，预计在八月亮相
• IT之家：OpenAI 预计 8 月初正式发布新一代 GPT-5 大模型

Qwen3-Coder成全球最热开源AI编程模型，阿里巴巴引领榜单

简报：

• 2025年7月24日，Hugging Face最新大模型排行榜显示，阿里巴巴Qwen3-Coder成功登顶，成为全球热度最高的开源AI编程模型，超越了GPT-4.1与Claude4等国际知名产品。
• Qwen3-Coder是千问系列首个采用混合专家（MoE）架构的代码模型，其在代码能力和Agent调用能力上表现突出，被业界公认为有望替代现有主流开源编程模型。
• 模型开源后下载量已超过2000万，受到全球开发者和AI社区的广泛关注，多位科技界领袖如Hugging Face首席执行官、Perplexity CEO等公开称赞。
• 此次Qwen3-Coder的登顶，标志着中国公司在全球AI开源模型领域的重大突破，为开发者提供更强编程支持，加速AI应用普及和创新。

相关链接：

• AI Base简报
• ITHome报道

Qwen3-Coder Becomes the World's Hottest Open-Source AI Coding Model, Alibaba Leads the Rankings

Briefing:

• July 24, 2025: According to Hugging Face's latest large model leaderboard, Alibaba's Qwen3-Coder successfully claimed the top spot 🏆, becoming the world's most popular open-source AI coding model, surpassing internationally renowned products like GPT-4.1 and Claude 4.
• Qwen3-Coder is the first code model in the Qwen series to adopt a Mixture-of-Experts (MoE) architecture. It has demonstrated outstanding performance in coding capabilities and Agent invocation, and is widely recognized by the industry as a potential successor to existing mainstream open-source programming models.
• Since its open-sourcing, the model has garnered over 20 million downloads, attracting widespread attention from global developers and the AI community. Several tech leaders, including the CEO of Hugging Face and the CEO of Perplexity, have publicly praised it. ✨
• Qwen3-Coder's ascent to the top marks a significant breakthrough for Chinese companies in the global open-source AI model domain, offering stronger programming support for developers and accelerating the popularization and innovation of AI applications. 🚀

Related Links:

• AI Base Briefing
• ITHome Report

百度AI搜索在IDC中国通用AI搜索评比中夺冠

简报：

• 国际数据公司（IDC）于2025年7月25日发布报告，评估中国通用型AI搜索产品，结果显示百度AI搜索总分排名第一。
• 在用户数据和技术能力两个核心指标方面，百度AI搜索均列榜首。
• 报告指出，百度凭借多年在搜索和大模型领域的技术积累，在意图识别准确率、搜索结果丰富度及精度等关键指标上获得满分，展现了其在多意图复杂查询中的稳定语义理解能力。

相关链接：

• 36氪快讯：IDC发布中国AI搜索产品评估 百度AI搜索排名第一

Baidu AI Search Tops IDC China's General AI Search Evaluation 🥇

Briefing:

• International Data Corporation (IDC) released a report on July 25, 2025, evaluating general-purpose AI search products in China, with Baidu AI Search ranking first in overall score. 🎉
• Baidu AI Search ranked first in both key indicators: user data and technical capability.
• The report highlighted that Baidu, leveraging its years of technical accumulation in search and large language models, achieved full marks in crucial metrics such as intent recognition accuracy, search result richness, and precision. This demonstrates its stable semantic understanding capability in multi-intent complex queries. 🚀

Related Link:

• 36Kr News Flash: IDC Releases China AI Search Product Evaluation, Baidu AI Search Ranks First

英特尔公布未来处理器和制程路线，18A工艺将支持至少三代产品

简报：

• 英特尔CEO陈立武在2025年第二季度财报电话会上确认，Nova Lake处理器计划于2026年底推出，服务器端Diamond Rapids预计2026年下半年，而Coral Rapids则暂定2028-2029年发布。
• Intel 18A制程节点将至少应用于未来三代客户端和服务器产品，后续14A节点将于2028-2029年推出，与台积电A14节奏类似。
• 英特尔将在未来数月公布其AI战略，并计划继续加强x86 CPU和XE GPU在AI领域的作用，同时对AI企业定制ASIC持开放态度。
• 数据中心团队将迎来新领导人，服务器处理器市场占比约为55%。

相关链接：

• 英特尔确认 Coral Rapids 处理器规划，Intel 18A 节点服务至少三代产品

Intel Unveils Future Processor and Process Roadmap, 18A Process to Support at Least Three Generations of Products 🚀

Briefing:

• Intel CEO Pat Gelsinger confirmed during the Q2 2025 earnings call that the Nova Lake processor is planned for launch by late 2026. Server-side Diamond Rapids is expected in the second half of 2026, while Coral Rapids is tentatively scheduled for release in 2028-2029. 🗓️
• The Intel 18A process node will be used for at least three future generations of client and server products. The subsequent 14A node is expected to launch in 2028-2029, similar to TSMC's A14 timeline.
• Intel will announce its AI strategy in the coming months and plans to continue strengthening the role of x86 CPUs and XE GPUs in the AI field, while remaining open to custom ASICs for enterprise AI. 💡
• The data center team will welcome new leadership. Server processor market share is approximately 55%.

Related Links:

• Intel Confirms Coral Rapids Processor Planning, Intel 18A Node to Serve At Least Three Generations of Products

蚂蚁集团设立通用人工智能研究中心，知名AI专家团队加盟

简报：

• 蚂蚁集团近日成立通用人工智能（AGI）研究中心，隶属于蚂蚁技术研究院，聚焦AGI基础技术创新。
• 该中心由蓝振忠（前谷歌科学家、ALBERT架构作者）担任主任，团队成员还包括武威、吴翼、赵俊博等多位AI领域新秀，他们具备丰富顶尖科研与产业化经验。
• 此次架构调整显示蚂蚁集团对基础AI研究的重视，意在推动通用人工智能领域的技术突破，而非单纯商业化探索。

相关链接：

• 蚂蚁集团新设通用人工智能研究中心，多位AI新秀加盟
• 蚂蚁集团人工智能企业服务总部、西部研发中心落地成都

Ant Group Establishes General Artificial Intelligence Research Center, Renowned AI Expert Team Joins

Brief:

• Ant Group recently established a General Artificial Intelligence (AGI) Research Center, affiliated with Ant Technology Institute, focusing on foundational AGI technology innovation.
• The center is led by Lan Zhenzhong (former Google scientist and author of the ALBERT architecture) as its director. The team also includes several rising stars in the AI field such as Wu Wei, Wu Yi, and Zhao Junbo, who possess extensive top-tier research and industrialization experience. 🚀
• This organizational restructuring demonstrates Ant Group's emphasis on foundational AI research, aiming to drive technological breakthroughs in the general artificial intelligence domain, rather than merely commercial exploration. ✨

Related Links:

• Ant Group Establishes New General Artificial Intelligence Research Center, Multiple AI Rising Stars Join
• Ant Group's AI Enterprise Service Headquarters and Western R&D Center Established in Chengdu 💡