![[1d74ec61-3742-427c-9cd2-e1b57300f247.mp3]]

Claude推出Artifacts功能，实现零代码AI应用创作与共享

简报：

• Anthropic公司宣布为其AI助手Claude推出全新Artifacts功能，允许用户无需编程知识，通过对话形式在Claude内快速创建、管理和分享个性化AI应用。
• 用户可在独立仪表盘上保存、组织、迭代作品，还可浏览并“Remix”他人应用，促进创意社区的共享与协作。
• Artifacts支持将Claude智能直接嵌入应用内，激励团队与个人共同创新，覆盖教育、数据分析、游戏等多元场景，极大降低了AI开发门槛。
• 每个Claude账户用户（含免费用户）均可访问Artifacts，企业团队还可在内部安全环境中共享应用，进一步提升商业与知识协作效率。

相关链接：

• AIbase新闻解读1
• AIbase新闻解读2

Claude Unveils Artifacts Feature: Zero-Code AI App Creation and Sharing 🚀

Briefing:

• Anthropic has announced the launch of a new Artifacts feature for its AI assistant, Claude, enabling users to quickly create, manage, and share personalized AI applications directly within Claude through conversational prompts, all without requiring programming knowledge.
• Users can save, organize, and iterate on their creations in a dedicated dashboard, and also browse and "Remix" applications from others, fostering a creative community for sharing and collaboration.
• Artifacts allows Claude's intelligence to be directly embedded within applications, encouraging teams and individuals to innovate together across diverse scenarios such as education, data analysis, and gaming, significantly lowering the barrier to AI development.
• All Claude account users (including free users) can access Artifacts, while enterprise teams can further enhance business and knowledge collaboration efficiency by sharing applications within a secure internal environment. 💡🤝

Related Links:

• AIbase News Interpretation 1
• AIbase News Interpretation 2

谷歌开源发布Gemini CLI，打造免费高性能AI终端助手

简报：

• 谷歌正式推出Gemini CLI，一款基于Gemini2.5Pro模型的开源AI代理工具，可在开发者命令行终端中免费使用AI能力。
• 工具支持100万token上下文窗口、每分钟60次和每日1000次免费请求，采用Apache2.0协议并集成Google搜索，实现实时信息抓取与自动化。
• Gemini CLI支持插件扩展、脚本自动化和VS Code协同，具备自动恢复、智能建议等编程辅助功能，还可调用模型生成视频和图像，满足多模态创作需求。
• 由于免费高额度和开源特点，Gemini CLI发布后在开发社区迅速走红，被视为对现有商业AI工具的强力挑战。

相关链接：

• AIBase报道：谷歌发布Gemini CLI：轻量开源AI代理免费赋能终端
• AIBase报道：Gemini CLI开源发布，免费提供AI编程助手挑战Cursor
• GitHub项目主页

Google Open-Sources Gemini CLI, Creating a Free High-Performance AI Terminal Assistant

Briefing:

• Google has officially launched Gemini CLI, an open-source AI agent tool based on the Gemini 2.5 Pro model, offering free AI capabilities directly within developers' command-line terminals. 🚀
• The tool supports a 1 million token context window, 60 free requests per minute, and 1000 free requests per day. It adopts the Apache 2.0 license and integrates Google Search for real-time information retrieval and automation.
• Gemini CLI supports plugin extensions, script automation, and VS Code collaboration. It features programming assistance functions like auto-recovery and intelligent suggestions, and can even invoke the model to generate videos and images, meeting diverse multimodal creation needs. 🎨
• Due to its generous free usage limits and open-source nature, Gemini CLI has quickly gained popularity within the developer community since its release, being seen as a powerful challenge to existing commercial AI tools. 💪

Related Links:

• AIBase Report: Google Releases Gemini CLI: Lightweight Open-Source AI Agent Empowers Terminals for Free
• AIBase Report: Gemini CLI Open-Sourced, Offering Free AI Programming Assistant Challenging Cursor
• GitHub Project Page

谷歌推出Imagen4与Imagen4 Ultra，推动文本生成图像技术升级

简报：

• 谷歌最新发布的文本生成图像模型Imagen4及其高精度版本Imagen4 Ultra，已通过Gemini API和Google AI Studio面向开发者开放付费预览和部分免费试用名额。
• Imagen4在文本渲染质量、生成速度和分辨率上较前代有显著提升，并支持多样化艺术风格创作；Ultra版本更适合对文本提示精准度有高要求的场景。
• 所有生成图像均嵌入SynthID数字水印以保障追溯性，谷歌还计划持续扩展模型可用性及优化定价，社区普遍认为其有望成为“文生图”领域新标杆。

相关链接：

• 谷歌发布Imagen4，文本到图像AI能力再次升级
• 谷歌推出Imagen4文生图AI，文本渲染及分辨率有明显提升

Google Launches Imagen4 and Imagen4 Ultra, Advancing Text-to-Image Technology Upgrade

Brief:

• Google's newly released text-to-image models, Imagen4 and its high-precision version Imagen4 Ultra, are now available to developers via Gemini API and Google AI Studio, offering paid preview and some free trial slots. 🚀
• Imagen4 boasts significant improvements over its predecessors in text rendering quality, generation speed, and resolution, also supporting diverse artistic style creation. The Ultra version is particularly suited for scenarios demanding high precision in text prompts. ✨
• All generated images are embedded with SynthID digital watermarks to ensure traceability. Google also plans to continuously expand model availability and optimize pricing, with the community widely anticipating it will set a new benchmark in the 'text-to-image' domain. 🎨

Related Links:

• Google Releases Imagen4, Text-to-Image AI Capabilities Level Up Again
• Google Launches Imagen4 Text-to-Image AI, with Significant Improvements in Text Rendering and Resolution

小米推出首款搭载相机与支付功能的AI眼镜

简报：

• 小米在全生态发布会上发布首款AI眼镜，重量40g，支持线下线上配镜及三种镜框选择。
• 眼镜搭载12MP第一人称相机，可拍照、录像，支持视频通话和直播，还配有开放式耳机和五麦克风系统。
• 内置“超级小爱”AI助手，支持语音操作、物体识别、文本翻译及扫码支付（需语音确认）。
• 该产品提供电致变色和彩色镜片版本，售价1999元起，配备独立Type-C接口支持边用边充电。

相关链接：

• IT之家：小米发布首款AI眼镜
• IT之家：小米AI眼镜详细规格

Xiaomi Launches First AI Glasses with Camera and Payment Features

Briefing:

• Xiaomi unveiled its first AI glasses at its all-scenario launch event, weighing 40g, supporting offline and online prescription lens fitting, and offering three frame choices. 👓
• The glasses feature a 12MP first-person camera for photos and videos, supporting video calls and live streaming. They also come with open-ear headphones and a five-microphone system. 📸
• It boasts a built-in "Super Xiao AI" assistant, enabling voice control, object recognition, text translation, and QR code payments (requires voice confirmation).
• The product is available in electrochromic and colored lens versions, starting from 1999 yuan. It features a dedicated Type-C port supporting charging while in use. 🚀

Related Links:

• IT Home: Xiaomi Launches First AI Glasses
• IT Home: Xiaomi AI Glasses Detailed Specifications

豆包AI以683分高考文科成绩领跑顶级模型，达清北录取线

简报：

• 字节跳动Seed团队公布2025高考全科测试成绩，豆包Seed1.6-Thinking模型在高考文科获得683分、理科648分，文科成绩为所有参测国内外顶级AI模型中最高，均已达到清华、北大录取分数线。
• 测试采用全国新一卷和山东省自主命题试卷，豆包在语文、英语、物理、历史、地理、政治六门学科均获最高单科分，并在数学中超过140分，展现出强大逻辑推理能力。
• 技术细节显示，高清试卷下豆包理科成绩可提升至676分，验证全模态推理提升AI综合能力，有力推动国产AI大模型在教育领域的深度应用。
• 资深高中教师估算，豆包赋分组合成绩最高可达690分，对应山东省排名前80，具备冲击顶级高校实力。

相关链接：

• AI 测高考丨豆包AI高考成绩达清北录取线 文科683分领跑
• IT之家相关报道
• 全模态推理技术驱动豆包AI高考成绩大幅提升

Doubao AI Leads Top Models with 683 Gaokao Liberal Arts Score, Reaching Tsinghua & Peking University Admission Line

Briefing:

• ByteDance's Seed team has released the 2025 Gaokao full-subject test results. The Doubao Seed1.6-Thinking model achieved 683 points in Liberal Arts and 648 points in Science, with its Liberal Arts score being the highest among all participating top domestic and international AI models. Both scores have reached the admission lines for Tsinghua University and Peking University. 🏆
• The test utilized the National New I Exam Paper and Shandong Province's self-developed exam papers. Doubao achieved the highest single-subject scores in six disciplines: Chinese, English, Physics, History, Geography, and Politics, and scored over 140 points in Mathematics, demonstrating its strong logical reasoning capabilities. 🤯
• Technical details show that Doubao's Science score could increase to 676 points under high-definition test papers, verifying that full-modality reasoning enhances AI's comprehensive abilities and strongly promotes the deep application of domestic AI large models in the education sector.
• Senior high school teachers estimate that Doubao's combined score could reach a maximum of 690 points, corresponding to a top 80 ranking in Shandong Province, indicating its potential to compete for admission to top-tier universities. ✨

Related Links:

• AI Gaokao Test | Doubao AI's Gaokao Score Reaches Tsinghua & Peking University Admission Line, Liberal Arts Leads with 683 Points
• IT Home Related Report
• Full-Modality Reasoning Technology Drives Significant Improvement in Doubao AI's Gaokao Performance