![[5d93e63a-2b71-49ef-bb22-d4c168d909ce.mp3]]

Meta AI应用上线强大视频编辑工具，可一键换装换景

简报：

• Meta Platforms正式在Meta AI应用、Meta.AI网站和Edits应用中推出AI驱动的视频编辑功能，允许用户使用超50种预设AI提示轻松替换视频中的服装、场景、光影等元素，并将编辑后的视频分享或发布。
• 该工具受Meta Movie Gen AI技术启发，可将人物瞬间转变为漫画、雕像或不同游戏角色，也能置于如海滩、雪景等虚拟场景下，目前暂不支持自定义提示词，官方称后续将开放该功能。
• 这是Meta Movie Gen AI首次商业化应用，未来还将逐步融合至Instagram等更多Meta平台产品。

相关链接：

• 36氪新闻快讯
• IT之家报道
• AIBASE资讯
• Meta官方新闻稿

Meta AI App Launches Powerful Video Editing Tool for One-Click Outfit and Scene Swaps ✨

Brief:

• Meta Platforms has officially launched AI-powered video editing features within the Meta AI app, Meta.AI website, and Edits app. This new tool allows users to easily replace elements like outfits, scenes, lighting, and more in videos using over 50 preset AI prompts, and to then share or publish their edited videos.
• Inspired by Meta's Movie Gen AI technology, this tool can instantly transform subjects into cartoons, statues, or different game characters, and place them into virtual environments like beaches or snowy landscapes. 🏖️ Currently, it does not support custom prompts, though the company states this functionality will be made available later.
• This marks the first commercial application of Meta Movie Gen AI, and it will gradually be integrated into more Meta platforms, including Instagram, in the future. 🚀

Related Links:

• 36kr news flash
• IT Home report
• AIBASE news
• Meta official press release

OpenAI大幅下调o3模型价格并推出高性能o3-pro新模型

简报：

• OpenAI宣布将o3旗舰推理模型的价格降低80%，输入令牌从每百万10美元降至2美元，输出令牌从每百万40美元降至8美元，缓存输入令牌享受额外75%折扣。
• 同时，OpenAI同步推出性能更强的o3-pro模型，主要面向高可靠性与复杂任务用户，定价为输入令牌每百万20美元、输出令牌每百万80美元，已在ChatGPT Pro及Team版上线。
• 本次调价及新品发布，被视为OpenAI对Google Gemini2.5Pro及Anthropic Claude4等竞品的直接回应，或引发AI行业新一轮价格战。
• 降价后，o3模型在代码、科研等领域的普及度有望进一步提升，o3-pro则为企业和教育用户提供更高性能选择。

相关链接：

• OpenAI launches o3-pro, slashes o3 price by 80% in bid to widen AI lead
• OpenAI announces 80% price drop for o3, it's most powerful reasoning model
• OpenAI adds o3 Pro to ChatGPT and drops o3 price by 80 per cent
• aibase快讯
• 36氪快讯
• aibase补充报道

OpenAI Drastically Cuts o3 Model Price and Introduces High-Performance o3-pro New Model 🎉

Briefing:

• OpenAI announced an 80% price reduction for its flagship o3 reasoning model. Input tokens decreased from $10 per million to$2 per million, and output tokens from $40 per million to$8 per million. Cached input tokens enjoy an additional 75% discount.
• Simultaneously, OpenAI launched the more powerful o3-pro model, primarily aimed at users requiring high reliability and complex tasks. Priced at $20 per million for input tokens and$80 per million for output tokens, it is now available on ChatGPT Pro and Team versions.
• This price adjustment and new product launch are seen as OpenAI's direct response to competitors like Google Gemini 2.5 Pro and Anthropic Claude 4, potentially triggering a new round of price wars in the AI industry. ⚔️
• Following the price drop, the o3 model's adoption in areas such as coding and scientific research is expected to further increase, while o3-pro offers a higher-performance option for enterprise and educational users. 🚀

Related Links:

• OpenAI launches o3-pro, slashes o3 price by 80% in bid to widen AI lead
• OpenAI announces 80% price drop for o3, it's most powerful reasoning model
• OpenAI adds o3 Pro to ChatGPT and drops o3 price by 80 per cent
• aibase快讯
• 36氪快讯
• aibase补充报道

DeepL借力英伟达新芯片大幅加速互联网翻译

简报：

• 德国翻译平台DeepL宣布启用英伟达最新DGX SuperPOD系统，称其可在18天内完成对整个互联网内容的翻译，远快于原需194天的速度。
• 新系统搭载36颗B200 Grace Blackwell Superchip，为DeepL训练和运行自有生成式AI模型提供算力支持，与谷歌翻译等产品展开竞争。
• DeepL希望通过提升算力，推动更先进AI模型的研发，升级后的硬件将强化如Clarify等新产品的能力，显著提升多语言翻译质量。

相关链接：

• IT之家报道

DeepL Drastically Accelerates Internet Translation with NVIDIA's New Chips 🚀

Briefing:

• German translation platform DeepL announced it has activated NVIDIA's latest DGX SuperPOD system, claiming it can now translate the entire internet's content in just 18 days, a drastic acceleration from the previously required 194 days. 💨
• The new system is equipped with 36 B200 Grace Blackwell Superchips, providing the computational power DeepL needs to train and run its proprietary generative AI models, enhancing its competition with products like Google Translate.
• DeepL aims to drive the development of more advanced AI models by significantly boosting its computing power. The upgraded hardware will enhance the capabilities of new products like Clarify, leading to a substantial improvement in multi-language translation quality. ✨

Related Link:

• IT Home Report

iOS 26升级快捷指令，Apple Intelligence深度整合AI提升用户体验

简报：

• 苹果在iOS 26、iPadOS 26及macOS Tahoe系统中对快捷指令（Shortcuts）应用进行重大升级，扩展多项Apple Intelligence（人工智能）功能。
• 新增技能涵盖PDF文档快速总结、根据食材生成食谱、自然语言查询、会议纪要提取行动事项、生成俳句与ASCII艺术、对比文档等多样化任务，支持端侧、私有云及ChatGPT等多种AI选项。
• 用户可自定义整合日历、天气、提醒事项等数据，实现高度个性化和复杂任务自动化，进一步提升日常操作的智能化和便捷性。
• 新系统同步强化隐私保护和AI能力，支持Visual Intelligence及Image Playground等创新功能，为用户带来全面升级的智能体验。

相关链接：

• 苹果 iOS 26 升级快捷指令：扩展引入 Apple 智能，AI 改变用户日常体验

iOS 26 Upgrades Shortcuts 🚀: Apple Intelligence Deeply Integrates AI to Enhance User Experience ✨

Brief:

• Apple is making significant upgrades to the Shortcuts app across iOS 26, iPadOS 26, and macOS Tahoe systems, expanding multiple Apple Intelligence (AI) features.
• New capabilities include quick PDF document summarization, recipe generation from ingredients, natural language queries, extracting action items from meeting minutes, generating haikus and ASCII art, document comparison, and other diverse tasks, supporting on-device, private cloud, and ChatGPT AI options. 🧠
• Users can customize data integration from Calendar, Weather, Reminders, and more, enabling highly personalized and complex task automation, further enhancing the intelligence and convenience of daily operations.
• The new systems simultaneously strengthen privacy protection and AI capabilities, supporting innovative features like Visual Intelligence and Image Playground, delivering a comprehensively upgraded intelligent experience for users.

Related Links:

• Apple iOS 26 Upgrades Shortcuts: Expands to Introduce Apple Intelligence, AI Changes User Daily Experience

火山引擎发布豆包1.6与Seedance1.0pro，AI模型性能提升且大幅降本

简报：

• 字节跳动在"Force原动力大会"上推出了火山引擎豆包大模型1.6及视频生成模型Seedance1.0pro，两者在多模态理解、视频生成等AI能力上取得重大突破。
• 豆包1.6在国际测评中表现优异，适用于电商识图、自动驾驶数据标注等多场景，操作能力大幅提升并可自动完成实际任务；其使用成本按输入长度定价，0-32K区间输入仅0.8元/百万tokens，综合成本为上一代的三分之一。
• Seedance1.0pro支持文字和图片输入生成1080P视频，在多项榜单成绩超越行业主流竞品，每条5秒1080P视频成本仅3.67元，极具性价比。
• 火山引擎还同步升级了AI云原生服务，推出新产品MCP服务和PromptPilot，助力AI应用与企业开发高效落地。

相关链接：

• 新闻链接1

VolcEngine Releases Doubao 1.6 and Seedance 1.0 Pro: Enhanced AI Model Performance and Significant Cost Reduction

Briefing:

• ByteDance unveiled VolcEngine's Doubao Large Model 1.6 and the video generation model Seedance 1.0 Pro at the "Force Yuan Dong Li Conference" (Force Original Power Conference). Both models have achieved significant breakthroughs in AI capabilities, including multimodal understanding and video generation. 🚀
• Doubao 1.6 has performed exceptionally well in international evaluations, suitable for various scenarios such as e-commerce image recognition and autonomous driving data labeling. Its operational capabilities have been greatly enhanced, allowing it to automatically complete practical tasks. Its usage cost is priced by input length, with the 0-32K token range costing only 0.8 yuan per million tokens, making the comprehensive cost one-third of the previous generation. 💰
• Seedance 1.0 Pro supports text and image input to generate 1080P videos, surpassing mainstream industry competitors in multiple rankings. The cost for each 5-second, 1080P video is only 3.67 yuan, offering extreme cost-effectiveness. 🎞️
• VolcEngine also synchronously upgraded its AI cloud-native services, introducing new products like MCP Service and PromptPilot, to help facilitate the efficient implementation of AI applications and enterprise development.

Related Links:

• News Link 1