AI NEWSLatest AI NewsArticleGemini 3.5 Pro Has Not Been Released Yet, Google May Skip the Wait and Turn to Gemini 4Published in Latest AI NewsTime :Aug 11, 2026Read :4minuteOn August 11, according to the latest report by industry analysis firm SemiAnalysis, Google may have put on hold its originally planned Gemini 3.5 Pro model and might be filling the market gap caused by the delay of its flagship model by promoting the Gemini 4 series in advance. Previously, Google had drawn attention due to the delayed release of Gemini 3.5 Pro, with reports indicating that the model still needs optimization in some key capabilities, especially in code-related performance.At the Google I/O developer conference in May 2026, Alphabet CEO Sundar Pichai stated that Gemini 3.5 Pro was expected to be launched in June, but the model has not officially been released yet. Subsequently, during Alphabet’s second-quarter earnings call, Pichai continued to emphasize Google’s progress in cutting-edge AI and expressed confidence in the development of the Gemini series models.SemiAnalysis pointed out that the current Gemini series is facing competitive pressure in some AI capability evaluations and believes its development speed lags behind some leading models. It was previously reported that Gemini 3.5 Pro briefly appeared on the Arena platform and was removed about 30 minutes after going online, sparking discussions about the model’s testing status.According to reports, Google’s delay in releasing Gemini 3.5 Pro was mainly related to model performance optimization, including enterprise-level application scenarios such as code generation. Google has stated that the model is still in the testing phase and is being evaluated with partners.If SemiAnalysis’s assessment proves correct, Google may be adjusting the Gemini product timeline, allocating more resources to the next-generation Gemini 4 series to cope with the accelerating model iteration from OpenAI, Anthropic, and other AI laboratories. Currently, competition among large models has shifted from simply launching new versions to a comprehensive contest involving capabilities, costs, application ecosystems, and R&D efficiency.

AI NEWSLatest AI NewsArticleNighttime Shock Release: DeepSeek V4 Pro Official Version Makes Its Debut, Standing Shoulder to Shoulder with Top Fable ModelsPublished in Latest AI NewsTime :Aug 13, 2026Read :2minuteOn August 13th, the AI field experienced a surge of product launches. While Musk launched Grok4.6 and Qwen3.8Max, an open-source model from Tongyi, DeepSeek also quietly released the official version of V4Pro - DeepSeek-V4-Pro-0813 at midnight.The official version is now fully available, with both its API and Chat interface updated. Users can access the latest version by simply specifying the relevant name without changing their existing methods.Official performance benchmark tests show that V4-Pro-0813 performs exceptionally well in all core metrics, not only surpassing Opus4.8 but also demonstrating strong competitiveness compared to the top-tier American Fable5 model, placing it in the first tier of the industry overall.In terms of pricing, this version continues to be three times the price of V4-Flash. Although the official had previously mentioned a pricing adjustment plan, the specific adjustment has not yet been displayed on the platform page, and it is expected that related price change notifications will be officially released through official channels later.

AI NEWSLatest AI NewsArticleClaude Fails to Solve a Century-Old Math Puzzle but Accidentally Rewrites the Fundamental Boundaries of Human Mathematical History!Published in Latest AI NewsTime :Aug 11, 2026Read :3minuteRecently, the field of artificial intelligence has witnessed a remarkable new breakthrough. Claude, an unreleased research version developed by Anthropic, attempted to challenge one of the highest peaks in the mathematical world, the Riemann Hypothesis. Although it did not fully solve this century-old problem, it unexpectedly increased the lower bound of the proportion of zeros on the critical line of the Riemann zeta function from 41.6%, which had been maintained for decades, to 67.2% during the exploration process.This achievement quickly sparked heated discussions in the academic community. It should be clarified that this does not mean the Riemann Hypothesis has been proven to 67.2%. The study does not provide definitive guidance for the hypothesis itself. However, its real significance lies in the new form of scientific research demonstrated by AI. During this multi-round effort, Claude did not rely on traditional search for known answers, but instead orchestrated approximately 60 sub-Agents to form a “virtual research team,” experiencing hundreds of failures, independently reviewing massive literature, and recombining classical methods. Ultimately, it explored innovative boundaries that had never been established before in the open and unknown field of mathematics.Industry experts point out that this breakthrough not only showcases the hard-core capabilities of artificial intelligence in complex, long-range, multi-agent collaboration, and autonomous error verification, but also signals a potential revolution in future research workflows. AI may no longer be limited to solving known problems, but instead, by undertaking intensive trial-and-error processes and cross-referencing vast amounts of literature, it will significantly reduce the marginal search costs for humans in unknown fields, ushering in a new era of scientific exploration.

AI NEWSLatest AI NewsArticleAI Copyright Crisis Escalates Again: Independent Music Publishers Sue Anthropic and Suno, Seeking Compensation Exceeding $1 BillionPublished in Latest AI NewsTime :Aug 18, 2026Read :3minuteCopyright lawsuits in the field of artificial intelligence are on the rise again. Independent music publisher Round Hill Music recently filed a lawsuit with the U.S. Federal Court in California, taking AI giants Anthropic and Suno to court, accusing both companies of unauthorized use of hundreds of copyrighted songs to train their generation systems.According to the complaint documents, Anthropic used at least 500 classic song lyrics from Round Hill’s copyright portfolio in the training process of its chatbot Claude, including works by famous musicians such as James Brown, The Kinks, and Goo Goo Dolls, to help the large model better understand and respond to various human prompts. At the same time, Round Hill also made the same allegations against Suno, pointing out that its AI music generation system also illegally used this copyright music library for training.Facing the escalating controversy over infringement, the music publisher revealed in both cases that, with further investigation and evidence collection, the number of involved songs is likely to expand to tens of thousands or more in the future. Based on this scale of potential infringement, the plaintiff is currently seeking a huge compensation of nearly or exceeding 1 billion U.S. dollars. As of now, neither Anthropic nor Suno has made an official public response to this lawsuit.

AI NEWSLatest AI NewsArticleKunlun Wanwei Launches Mureka V9.5: AI Music Moves Away from Mechanical Arrangement and Toward Human-Hearted CreationPublished in Latest AI NewsTime :Aug 18, 2026Read :5minuteKunlun Wanyi officially launched its new AI music generation model, Mureka V9.5 today. This version abandons the current industry’s common “high volume, high saturation, strong stimulation” competition path, and instead emphasizes “restraint, blank space, and naturalness,” aiming to solve the long-standing “mechanical feel” problem in AI music, making the generated music truly have a “human touch” and stand up to repeated listening.The core upgrade of Mureka V9.5 lies in the deepening of its self-developed MusiCoT (Music Chain of Thought) framework. Unlike previous AI that directly generates audio waveforms, V9.5 first builds a strict “music thought process” before generating, planning the complete logic from macro musical structure to micro note expression. This allows the model to “do more with less”: arrangements no longer constantly overload, but leave “breathing space” for melody and emotion; the evolution of sections is more in line with human creative logic, significantly reducing template stacking.According to internal evaluation data, V9.5 has achieved significant breakthroughs in key indicators: the pass rate of vocal performance has risen to 61.0%, and the expression of breath and emotion is closer to that of real people; the pass rate of prompt control is as high as 97.0%, and the proportion of completely reflecting the style reaches 95.7%, meaning the model can better execute the creator’s intention, rather than covering up logical deficiencies with complex instrumentation.Notably, V9.5’s performance in the Chinese traditional music field is truly “transformative.” The new model has greatly improved the accuracy of Chinese pronunciation, solving the problem of unclear long sentences; at the same time, the handling of breath and emotional fluctuations is more natural, completely leaving behind the “synthetic feeling” commonly found in traditional Chinese AI music. In addition, the model has optimized the balance between harmony and melody, accurately restoring the essence of national instrumentation, and refusing to simply pile up “Chinese-style” elements.Kunlun Wanyi stated that the evolutionary direction of V9.5 directly comes from over 25,000 genuine user feedback from around the world. User needs clearly point toward “stronger consistency in long passages, more realistic vocals, and music that truly sounds human.” This marks that Mureka is evolving from a mere “generation tool” into an inspiration catalyst and a part of professional workflow.Looking back at Mureka’s development: the V8 version achieved “AI music can be released,” the V9 version achieved “fine control,” while V9.5 has given AI music the “soul” to be worth listening to repeatedly. Currently, Mureka V9.5 is open to global users. Overseas users can visit mureka.ai, and domestic users can visit mureka.cn to experience it.

AI NEWSLatest AI NewsArticleAnthropic Reveals Model2 for the First Time: Performance Exceeds Current Strongest AI, Further Expanding the Boundaries of TechnologyPublished in Latest AI NewsTime :Aug 18, 2026Read :2minuteRecent developments in the field of artificial intelligence have emerged. According to the “2026 August Risk Report” released by Anthropic on August 15, the company has for the first time publicly disclosed a new AI model that has not yet been officially released - Model 2.From the details revealed in the report, Model 2 shows superior overall performance compared to the currently strongest publicly available model, Claude Mythos 5, on multiple benchmark tasks. However, industry analysts point out that the performance improvement brought by Model 2 compared to its predecessor is relatively moderate. Its core significance lies in further consolidating and enhancing the overall capabilities of AI models, rather than making a disruptive leap.Additionally, according to the internal “CoBench v2” benchmark test results, the report also included related terminology about Model 1. The outside world infers that Model 2 is likely a successor to Model 1, and the two are probably further deep iterations and upgrades under the Claude Mythos Preview technology roadmap.

AI NEWSLatest AI NewsArticleOpenRouter Announces 50% Reduction in GPT-5.6Sol API Call CostsPublished in Latest AI NewsTime :Aug 18, 2026Read :2minuteGlobal AI model aggregation platform OpenRouter announced on August 17 on the X platform that the call fee for the OpenAI GPT-5.6Sol model has been reduced by 50%. After the adjustment, the input price per 1 million Tokens was reduced from $5 to $2.5, and the output price from $30 to $15, which is approximately 16.9 yuan and 101.3 yuan in RMB, respectively.In addition to OpenRouter, the AI hosting platform Vercel also announced a limited-time half-price offer, valid for one month until September 18, 2026, applicable to the standard and Fast modes of GPT-5.6Sol. Currently, OpenRouter has not yet announced an exact time limit for the promotion.Industry analysis media SemiAnalysis believes that this price cut may have strong promotional and market competition intentions. OpenRouter and Vercel are both important platforms for AI model evaluation, development, and invocation. Lowering the usage cost of GPT-5.6Sol is expected to attract more developers and users to perform actual invocations, thereby expanding the model’s usage scale and related evaluation data.

ENLatest AI NewsShattering Perceptions! Ultraman Says Four Years of College Is Too Long - How to Redefine Personal Growth and Career Paths in the AI Era?OpenAI CEO Sam Altman said at Internapalooza that four-year undergraduate degrees take too long in the AI era, and two years of college is just right, sparking debate on how traditional higher education and career planning should adapt to AI disruption…..just now5.6KHalf price! OpenRouter lowers the call fee for OpenAI’s strongest model GPT-5.6 SolOpenRouter announced that it will halve the call fee for OpenAI’s GPT-5.6 Sol: the input cost per million Tokens will drop from $5 to $2.5, and the output cost from $30 to $15, equivalent to approximately RMB 16.9 yuan and 101.3 yuan, respectively. It has not yet been specified whether this is a time-limited offer. Vercel also launched a limited-time half-price promotion, valid for one month until September 18, 2026, covering standard and Fast modes.just now5.5KBrockman Sounds the Alarm: AI Breaks Through Hugging Face, a Milestone - Enterprises Should Immediately Do 10 ThingsOpenAI President Brockman warned its AI model breached Hugging Face, a cybersecurity watershed. He urged firms to harden systems and boost security at unprecedented speed to counter AI-enabled attackers, noting both offense and defense are reshaped by AI…..just now3.5KOpenRouter Announces 50% Reduction in GPT-5.6Sol API Call CostsOpenRouter announced a 50% reduction in the API call costs for the OpenAI GPT-5.6Sol model: the input cost per million Tokens has been reduced from $5 to $2.5, and the output cost from $30 to $15. AI hosting platform Vercel also launched a limited-time half-price offer, valid until September 18, 2026, covering standard and Fast modes; OpenRouter has not yet announced the duration of the promotion.just now7.0KVercel Open Sources Zero Language: Designed for AI Agents, Treat Graphs as Source CodeVercel Labs open-sources the experimental language Zero, aimed at AI agents rather than humans, using semantic graphs instead of text as the source; agents query the program graph and submit validated patches, which the compiler accepts or rejects directly, with humans only occasionally viewing the text projection, achieving the compiler inside the loop and humans outside. The project has already received 5310 stars.just now5.5KAnthropic Reveals Model2 for the First Time: Performance Exceeds Current Strongest AI, Further Expanding the Boundaries of TechnologyAnthropic released the ‘Risk Report August 2026’ on August 15, revealing its new unannounced model, Model 2. The model shows superior performance in multiple benchmark tests compared to the current strongest industry model, Claude Mythos 5, although the performance improvement is relatively moderate. Industry experts believe its core significance lies in further strengthening the foundation.just now6.2KKunlun Wanwei Launches Mureka V9.5: AI Music Moves Away from Mechanical Arrangement and Toward Human-Hearted CreationKunlun Wanwei launched AI music model Mureka V9.5, shifting from loud, saturated stimulation to restraint, space, and naturalness to remove mechanical feel and add human warmth. Its core upgrade deepens the MusiCoT music chain-of-thought framework, moving beyond direct waveform generation for more natural, listenable music…..just now7.5KAI Copyright Crisis Escalates Again: Independent Music Publishers Sue Anthropic and Suno, Seeking Compensation Exceeding $1 BillionRound Hill Music sued Anthropic and Suno in California federal court, alleging unauthorized use of hundreds of copyrighted songs to train AI. It says Claude used lyrics from at least 500 Round Hill songs, including James Brown and The Kinks classics, highlighting rising AI copyright litigation…..just now7.8KLeopard Bird iO Series to be Released on August 21, Focused on Human Augmentation AI GlassesLeopard Bird Innovation announced on August 18 that the first product of the new Leopard Bird iO Series, the iO AI Glasses, will be released on August 21 at 14:30. The series name “iO” is derived from computer I/O, positioning it as the next-generation personal smart terminal. It aims to establish a more natural interaction between humans and the digital world through glasses. The official call it “Human Augmentation AI Glasses,” which not only provides information but also helps people see more, remember better, and express themselves more effectively, making AI capabilities “grow” onto humans.just now7.2KProduction-Grade RAG and Agent Visualization Platform Ollmo v0.1.0 Officially ReleasedOllmo v0.1.0 is a production-grade visual RAG and Agent platform for private deployment and team collaboration. Users upload documents to ask questions with precise citations, quickly building private knowledge bases. Backend: Go + Fiber; frontend: Next.js + shadcn-ui. Docker one-click deployment; supports private delivery and SaaS…..just now8.2KInside the Google AI Restructuring: Brin Takes Personal Control Over

感谢IT之家网友shutiao、木木人工、Kazuto、Domado、浪花庆山粉、AlanMac、华南吴彦祖、LL J、乌蝇哥的左手、files、火瓶座的线索投递！IT之家8 月 14 日消息，当地时间 8 月 13 日，谷歌宣布推出 Gemini 3.7 Flash 模型，专为编程与 Agents 场景打造。IT之家注意到，本次发布距 Gemini 3.6 Flash 发布仅间隔约三周，谷歌称该版本的推出直接源于开发者的反馈以及一系列算法创新。Gemini 3.7 Flash 在软件工程、知识工作及网页开发工作流方面均实现了大幅提升。此外，在今年年底前（截至 2026 年 12 月 31 日），Gemini 3.7 Flash 将以首发优惠价提供（Gemini 3.6 Flash 也将有相同优惠），仅为原 3.6 Flash 每百万 Token 价格的一半：输入：每百万 Token 0.75 美元（现汇率约合 5.1 元人民币）输出：每百万 Token 3.75 美元（现汇率约合 25.3 元人民币）即日起，面向 160 多个国家和地区的 Google AI Pro 及 Ultra 订阅用户推出的 Gemini Spark 将正式接入 Gemini 3.7 Flash。本次模型升级显著提升了 Spark 处理知识工作的效率，优化了其对 Google Workspace 系列应用工具的使用能力，从而为复杂且跨技能的综合工作流提供更高的准确度与更高质量的输出。此外，Gemini 3.7 Flash 推出了全新的安全防御机制，在严防化学、生物、放射性及核（CBRN）领域和网络攻击滥用的同时，确保合法且有益的应用场景得以顺利实施。

Grok Bot – SpaceXAI 推出的 AI Agent 系统，多 Bot 并行协作AI工具5天前发布AI小集05Grok Bot是什么Grok Bot 是 SpaceXAI 推出的 AI Agent 系统，能登录用户的账号像真人一样操作网站和工具，24 小时在云端替用户完成实际工作。系统支持多 Bot 并行协作、录制工作流自动执行，甚至 Bot 可创建新 Bot 组建团队。原型来自 Cursor 内部代号「Sand」的 Agent 产品，现架在 Cursor 基础设施上。Grok Bot的主要功能账号代操作：直接登录已有的应用和网站，在界面层像真人一样点击、填写、推进任务，无需 API 或 MCP 支持。云端常驻运行：每个 Bot 拥有专属小型云计算机，24 小时后台运行，不依赖你的设备是否开机，可凌晨自动拉取数据、整理报告。多 Bot 并行协作：可同时创建销售 Bot、招聘 Bot、费用 Bot、Bug 复现 Bot 等，各自独立处理不同任务。Bot 间自主通信：Bot 之间可直接对话、传递任务、协商分工；管理 Bot 可汇总子 Bot 进展，只在关键节点找你确认。Bot 创建 Bot：可指令 Bot 自行”组建团队”，自动创建新的子 Bot 并分配职责。工作流录制与演化：观察你的操作过程并保存为模板，支持在执行中纠正，工作流会随使用不断优化。跨端同步：支持 macOS 桌面端和 iOS，任务进度多端同步。如何使用Grok Bot访问下载：访问Grok Bot官网 https://x.ai/bot，下载 macOS 或 iOS 客户端。登录订阅：用已订阅 Cursor Teams Premium（$120/月）、Cursor Ultra（$200/月）或 SuperGrok Heavy（$300/月）的账号登录。创建 Bot：点击创建新 Bot，为其命名、设置职位头衔和任务描述。分配任务：给 Bot 指定具体工作，如销售外联、招聘筛选、费用管理或 Bug 复现。自动执行：Bot 自动登录指定的应用或网站，像真人一样在界面层操作并自主推进任务。人机协作：Bot 在云端 24 小时运行，仅在需要人类判断的关键节点回来找你确认。多 Bot 协同：可同时创建多个 Bot 并行运行，它们会互相通信、协商分工，甚至自动创建子 Bot 组建团队。Grok Bot的核心优势无接口限制：无需 API 或 MCP，直接登录应用和网站，在界面层像真人一样点击、填写、推进任务。云端常驻：每个 Bot 拥有专属云计算机，24 小时后台独立运行，不依赖你的设备是否在线。多 Bot 协同：可同时部署多个 Bot 处理不同任务，它们之间自主通信、协商分工、协同推进复杂工作流。工作流演化：观察并保存你的操作步骤为模板，支持执行中纠正，工作流随使用持续优化。自组织团队：可指令 Bot 自行组建团队，自动创建子 Bot 并分配职责，实现层级化管理。闭环执行：Grok Bot 能推进到任务最后一步，只在关键节点找你确认。Grok Bot的同类竞品对比对比维度Grok BotOpenClaw运行模式云端常驻，每个 Bot 拥有专属小型云计算机，24 小时后台独立运行，不依赖用户设备在线。本地优先，任务在用户本地设备或自托管服务器上执行，设备关机则任务中断。部署架构中心化托管，Bot 在 xAI 云端运行，用户通过客户端远程查看进度和接收通知。去中心化自托管，数据不经过第三方服务器，支持 Docker 一键部署，隐私可控性更强。交互入口专用桌面客户端（macOS）+ iOS App，Bot 以独立聊天窗口形式存在，支持多 Bot 并行会话。支持 Telegram、Discord、Slack 等 50 多个消息平台接入，通过聊天指令触发任务。模型支持深度绑定 Grok 系列模型，底层由 xAI 自研模型驱动，模型能力统一但不可切换。可自由切换 OpenAI、Anthropic、Google 及本地开源模型，用户可按任务需求灵活配置。工具操作直接登录用户账号，在应用界面层像真人一样点击、填写、操作，无需 API 或 MCP 支持。依赖 MCP（Model Context Protocol）和 API 连接工具，对无接口的传统软件支持有限。多 Agent 协同原生支持多 Bot 并行运行，Bot 之间可直接通信、协商分工、自动创建子 Bot 组建团队。单次会话为主，不同任务间缺乏持久化协同，多 Agent 协作需手动串联。工作流能力支持录制用户操作并保存为模板，Bot 可自主按模板执行并随纠正不断优化，实现工作流演化。通过预设指令和脚本定义工作流，缺乏”观察-学习-自动优化”的闭环演化机制。安全与隐私账号凭证托管在 xAI 云端，Bot 代登录操作，用户需信任平台的数据安全策略。本地运行，账号凭证和任务数据保留在用户侧，不经过第三方服务器，隐私可控性更高。定价门槛需订阅 Cursor Teams Premium（$120/月）、Cursor Ultra（$200/月）或 SuperGrok Heavy（$300/月）。开源免费，仅需承担自托管的服务器成本，对个人开发者和小团队门槛极低。Grok Bot的应用场景销售外联：夜间自动研究客户账户、拉取 Salesforce 数据、起草个性化邮件，早上留给你审批后发送。招聘筛选：自动浏览 LinkedIn 简历、匹配岗位需求、生成候选人短名单并安排面试。费用管理：从 Gmail 提取收据、分类归档、核对金额，定期生成报销报告。Bug 复现：自动登录测试环境、按步骤复现用户提交的 Bug、提交工单并移交给调试 Bot 跟进。内容运营：每小时扫描 Slack 频道查找会议信息，提醒相关人员并提供社交用语建议。# AI工具# AI项目和框架©版权声明本站文章版权归AI工具集所有，未经允许禁止任何形式的转载。上一篇ClawTeams - AI 电商运营协同平台，专家 Agent 并行执行下一篇dots.tts - 小红书联合上海交大开源的语音合成基座模型相关文章PixelRAG – 伯克利开源的视觉原生 RAG 框架AI小集3Stickerbox – AI创意贴纸打印机，语音交互生成贴纸AI小集3EvoX – EvoMap 推出的自进化 AI Agent，首创蜂群模式AI小集3MeshifAI – AI 3D模型生成平台，支持生成简单模型和纹理模型AI小集2Kiss3DGen – 基于图像扩散模型的3D资产生成框架AI小集2Nerd AI – AI学习应用，解答各类数学问题AI小集3暂无评论再想想发表评论暂无评论…热门工具Loomy即梦AISekoAiPPT秘塔AI搜索妙呀堆友Agent美图设计室绘蛙AIupdream办公小浣熊秒哒最新收录FloatboatPreviewOnSoloEpixa知漫剧QwenPaw最新文章秒悟Meoo – 1分钟生成前后端完整的网站，真0门槛开发！3分钟前Mureka V9.5 – 昆仑万维推出的新一代 AI 音乐生成模型3分钟前Luna-TTS – 宇生月伴推出的文本转语音大模型4分钟前MyContext – 阿里千问办公开源的 Agent 上下文基础设施5分钟前用 AI 做歌一直火不起来？问题不在灵感，在你没有制作人思维6分钟前Origin – Cursor 推出的 AI 原生代码托管平台5小时前if Studio – 阿里 iconfont 推出的企业级 AI 智能体工作台5小时前完全不懂乐理，我用AI做出了第一首能上线的原创歌5小时前HiDream-O1-World – 智象未来推出的全模态交互式世界模型21小时前DeepSeek Harness Desktop – 开源的 DeepSeek Harness 桌面端21小时前dsh-TUI – 开源的 DeepSeek Harness 终端界面插件1天前DSH Desktop – 开源的 DeepSeek Harness 本地桌面应用1天前WorkSwarm – openJiuwen 推出的蜂群办公智能体2天前Happy Shrimp – 阿里 ATH 事业群推出的 AI 音乐平台2天前Copula Lab – AI 数据基础设施平台，提供专家级训练数据2天前