[{"data":1,"prerenderedAt":1281},["ShallowReactive",2],{"tools-in-image":3,"news-for-image":1112},[4,113,232,305,508,690,768,841,917],{"id":5,"title":6,"body":7,"category":92,"description":13,"draft":93,"experience":93,"extension":94,"hasApi":95,"meta":96,"name":97,"navigation":95,"officialUrl":98,"path":99,"pricing":100,"seo":101,"slug":102,"stem":103,"tagline":104,"tags":105,"updatedAt":110,"vendor":111,"__hash__":112},"tools\u002Ftools\u002Fimage\u002Fflux.md","Flux",{"type":8,"value":9,"toc":79},"minimark",[10,14,19,25,38,41,44,47,50,53,61,64,69,72,76],[11,12,13],"p",{},"FLUX 是 Black Forest Labs 的图像生成模型系列，既有可本地部署的开源权重，也有官方 API。相比闭源模型，它的优势是能自己跑、能微调、能接进 ComfyUI 这类工作流。",[15,16,18],"h2",{"id":17},"_2026-年现状","2026 年现状",[20,21,22],"blockquote",{},[11,23,24],{},"数据核对至 2026-11-06。本站未实测该工具（experience: false），价格与功能以官网当前页面为准。",[26,27,28,32,35],"ul",{},[29,30,31],"li",{},"GitHub 上 black-forest-labs\u002Fflux 约 26,000 stars（GitHub API，2026-11 抓取）。",[29,33,34],{},"官方提供 FLUX 系列推理仓库与 BFL API，API 按图片尺寸与模型计费，官网公示单张价格约 0.01–0.60 美元区间。",[29,36,37],{},"开源权重可本地部署，商用授权与自托管条款以官网为准。",[15,39,40],{"id":40},"能干什么",[11,42,43],{},"FLUX 是 Black Forest Labs 的图像生成模型系列，既有可本地部署的开源权重，也有官方 API。相比闭源模型，它的优势是能自己跑、能微调、能接进 ComfyUI 这类工作流。 具体功能边界以官网文档为准，本站只列出能确认的能力，不替厂商写宣传语。",[15,45,46],{"id":46},"价格",[11,48,49],{},"开源权重可本地部署；API 按张计费（以官网为准）。价格和优惠随时可能调整，下单前请以官网结账页为准。",[15,51,52],{"id":52},"适合谁",[26,54,55,58],{},[29,56,57],{},"需要这类能力、并且愿意按官方页面自行核对价格的用户。",[29,59,60],{},"不适合：只想要完全免费、不接受学习成本的人。",[15,62,63],{"id":63},"常见问题",[65,66,68],"h3",{"id":67},"flux-是免费的吗","FLUX 是免费的吗？",[11,70,71],{},"开源权重可本地部署；API 按张计费（以官网为准）。免费额度和计费方式以官网为准。",[65,73,75],{"id":74},"这份资料是什么时候核对的","这份资料是什么时候核对的？",[11,77,78],{},"2026-11-06。本站标注了「未实测」，没有亲自付费使用过的工具不会写使用体验。",{"title":80,"searchDepth":81,"depth":81,"links":82},"",2,[83,84,85,86,87],{"id":17,"depth":81,"text":18},{"id":40,"depth":81,"text":40},{"id":46,"depth":81,"text":46},{"id":52,"depth":81,"text":52},{"id":63,"depth":81,"text":63,"children":88},[89,91],{"id":67,"depth":90,"text":68},3,{"id":74,"depth":90,"text":75},"image",false,"md",true,{},"FLUX","https:\u002F\u002Fbfl.ai","\u002Ftools\u002Fimage\u002Fflux","开源权重可本地部署；API 按张计费（以官网为准）",{"description":13},"flux","tools\u002Fimage\u002Fflux","Black Forest Labs 的开源图像模型，可本地部署、可商用授权，社区生态大",[106,107,108,109],"免费","付费","国外","有API","2026-11-06","Black Forest Labs","0XXfhnX-FOT6xsQPX8MYarnVJZq9CqY7x0e7W4v3xIY",{"id":114,"title":115,"body":116,"category":92,"description":120,"draft":93,"experience":93,"extension":94,"hasApi":95,"meta":220,"name":221,"navigation":95,"officialUrl":222,"path":223,"pricing":224,"seo":225,"slug":226,"stem":227,"tagline":228,"tags":229,"updatedAt":110,"vendor":230,"__hash__":231},"tools\u002Ftools\u002Fimage\u002Fgpt-image.md","Gpt Image",{"type":8,"value":117,"toc":209},[118,121,123,128,139,141,168,170,173,175,186,188,192,195,199,202,206],[11,119,120],{},"GPT Image 是 OpenAI 的图像生成模型。DALL·E 3 在 2023 年推出后，OpenAI 把图像生成并入 ChatGPT，模型线在 2026 年更新到 Images 2.6。对普通用户来说，入口还是 ChatGPT 里直接说「画一张」。",[15,122,18],{"id":17},[20,124,125],{},[11,126,127],{},"数据核对至 2026-11-06。价格与功能以官网为准。",[26,129,130,133,136],{},[29,131,132],{},"OpenAI 2026 年 11 月上线 Images 2.6，ChatGPT 内可直接调用（OpenAI News RSS）。",[29,134,135],{},"据量子位 2026-11-05 报道，GPT-Image-2.6 全量上线，支持直出 GIF 和 SVG。",[29,137,138],{},"API 按图像数量和尺寸计费，具体价格以 OpenAI 定价页为准。",[15,140,40],{"id":40},[26,142,143,150,156,162],{},[29,144,145,149],{},[146,147,148],"strong",{},"文生图","：在 ChatGPT 里用自然语言描述画面，不需要学 prompt 参数。",[29,151,152,155],{},[146,153,154],{},"文字渲染","：英文招牌、标题字这类需求，出图时文字准确率比早期模型高。",[29,157,158,161],{},[146,159,160],{},"多轮修改","：基于上一张图对话式调整背景、构图、风格。",[29,163,164,167],{},[146,165,166],{},"GIF \u002F SVG","：Images 2.6 开始支持直接生成 GIF 和 SVG 这类格式。",[15,169,46],{"id":46},[11,171,172],{},"图片生成包含在 ChatGPT 付费方案中，API 按生成张数与尺寸计费。以 OpenAI 定价页为准。",[15,174,52],{"id":52},[26,176,177,180,183],{},[29,178,179],{},"已经在用 ChatGPT 付费版、需要文章配图和封面的内容团队。",[29,181,182],{},"需要快速做概念图、海报草图的人。",[29,184,185],{},"不适合：要极致画质和艺术感的商业摄影级出图——这类需求 Midjourney、FLUX 更合适。",[15,187,63],{"id":63},[65,189,191],{"id":190},"dalle-3-还能用吗","DALL·E 3 还能用吗？",[11,193,194],{},"DALL·E 3 是 2023 年的模型，2026 年 OpenAI 的图像生成已更新到 Images 2.6，入口整合在 ChatGPT 里。老模型名在 API 侧可能仍可用，具体以 OpenAI 文档为准。",[65,196,198],{"id":197},"gpt-image-26-能生成中文文字吗","GPT Image 2.6 能生成中文文字吗？",[11,200,201],{},"英文文字渲染相对可靠，中文复杂字形仍容易出错，建议后期用设计软件补字。",[65,203,205],{"id":204},"价格怎么算","价格怎么算？",[11,207,208],{},"含在 ChatGPT 付费方案里；API 按生成张数与尺寸计费，以 OpenAI 定价页为准。",{"title":80,"searchDepth":81,"depth":81,"links":210},[211,212,213,214,215],{"id":17,"depth":81,"text":18},{"id":40,"depth":81,"text":40},{"id":46,"depth":81,"text":46},{"id":52,"depth":81,"text":52},{"id":63,"depth":81,"text":63,"children":216},[217,218,219],{"id":190,"depth":90,"text":191},{"id":197,"depth":90,"text":198},{"id":204,"depth":90,"text":205},{},"GPT Image 2.6","https:\u002F\u002Fopenai.com\u002Findex\u002Fintroducing-images-2-6\u002F","\u002Ftools\u002Fimage\u002Fgpt-image","含在 ChatGPT 付费方案中，API 按量计费（以官网为准）",{"description":120},"gpt-image","tools\u002Fimage\u002Fgpt-image","OpenAI 的图像生成模型，2026 年 11 月更新到 Images 2.6，支持直出 GIF 和 SVG",[107,108,109],"OpenAI","R4y0PNMpOB1TmkXVXVR5JADcIchQBHuhQKY2g7QFpVA",{"id":233,"title":234,"body":235,"category":92,"description":239,"draft":93,"experience":93,"extension":94,"hasApi":93,"meta":295,"name":234,"navigation":95,"officialUrl":296,"path":297,"pricing":298,"seo":299,"slug":300,"stem":301,"tagline":302,"tags":303,"updatedAt":110,"vendor":234,"__hash__":304},"tools\u002Ftools\u002Fimage\u002Fideogram.md","Ideogram",{"type":8,"value":236,"toc":285},[237,240,242,246,254,256,259,261,264,266,272,274,278,281,283],[11,238,239],{},"Ideogram 是主打文字渲染的 AI 图像生成工具，能在图片里较准确地排英文文字，适合做海报、社媒封面和 Logo 草图。官网 2026 年定价页有反爬保护，具体档位与价格以官网为准。",[15,241,18],{"id":17},[20,243,244],{},[11,245,24],{},[26,247,248,251],{},[29,249,250],{},"Ideogram 官网 2026 年仍以文字渲染为主打，提供免费额度和订阅档；具体价格以官网为准。",[29,252,253],{},"官网有反爬保护，本站未能抓取实时定价，价格与额度请以结账页为准。",[15,255,40],{"id":40},[11,257,258],{},"Ideogram 是主打文字渲染的 AI 图像生成工具，能在图片里较准确地排英文文字，适合做海报、社媒封面和 Logo 草图。官网 2026 年定价页有反爬保护，具体档位与价格以官网为准。 具体功能边界以官网文档为准，本站只列出能确认的能力，不替厂商写宣传语。",[15,260,46],{"id":46},[11,262,263],{},"免费版 + 订阅（价格以官网为准）。价格和优惠随时可能调整，下单前请以官网结账页为准。",[15,265,52],{"id":52},[26,267,268,270],{},[29,269,57],{},[29,271,60],{},[15,273,63],{"id":63},[65,275,277],{"id":276},"ideogram-是免费的吗","Ideogram 是免费的吗？",[11,279,280],{},"免费版 + 订阅（价格以官网为准）。免费额度和计费方式以官网为准。",[65,282,75],{"id":74},[11,284,78],{},{"title":80,"searchDepth":81,"depth":81,"links":286},[287,288,289,290,291],{"id":17,"depth":81,"text":18},{"id":40,"depth":81,"text":40},{"id":46,"depth":81,"text":46},{"id":52,"depth":81,"text":52},{"id":63,"depth":81,"text":63,"children":292},[293,294],{"id":276,"depth":90,"text":277},{"id":74,"depth":90,"text":75},{},"https:\u002F\u002Fideogram.ai","\u002Ftools\u002Fimage\u002Fideogram","免费版 + 订阅（价格以官网为准）",{"description":239},"ideogram","tools\u002Fimage\u002Fideogram","以文字渲染见长的 AI 图像工具，适合做海报、Logo 和带文字的视觉",[106,107,108],"sF1j3z08mcsh9DbYg1q7chA3UaFb2SgXyDpO1AV9GW0",{"id":306,"title":307,"body":308,"category":92,"description":312,"draft":93,"experience":95,"extension":94,"hasApi":93,"meta":495,"name":496,"navigation":95,"officialUrl":497,"path":498,"pricing":499,"seo":500,"slug":501,"stem":502,"tagline":503,"tags":504,"updatedAt":110,"vendor":506,"__hash__":507},"tools\u002Ftools\u002Fimage\u002Fjimeng.md","Jimeng",{"type":8,"value":309,"toc":481},[310,313,315,320,331,336,338,343,349,355,361,363,366,369,372,375,378,381,384,387,390,396,413,418,435,437,451,453,457,460,464,467,471,474,478],[11,311,312],{},"即梦是字节跳动在 2024 年推出的 AI 绘画工具,主要能做根据文字描述生成图像和视频。它的核心卖点是中文 prompt 理解好、国内直接用、免费额度多。出图质量不如 Midjourney,但在国内免费工具里算好的。和剪映打通,生成的图可以直接拿去剪映做视频。",[15,314,18],{"id":17},[20,316,317],{},[11,318,319],{},"数据核对至 2026-11-06。价格与功能以官网为准，核实时以官方页面为准。",[26,321,322,325,328],{},[29,323,324],{},"即梦 2026 年使用 Seedance 2.0 \u002F 2.5 视频模型，支持文生视频、图生视频与智能画布。",[29,326,327],{},"官网 2026 年价格：基础会员首年 ¥393（原价 ¥659\u002F年，725 积分\u002F月）、标准会员 ¥1099\u002F年、高级会员 ¥5199\u002F年（12,320 积分\u002F月）。",[29,329,330],{},"积分按生成时长消耗，Seedance 2.5 720P 限时约 ¥0.42\u002F秒，活动价以官网为准。",[20,332,333],{},[11,334,335],{},"说明：以下「能干什么 \u002F 使用经验 \u002F 优缺点」保留了我们在 2024–2025 年实际使用时的观察，当时的模型版本与界面与今天不同；2026 年的产品现状见上方「2026 年现状」，价格与功能以官网为准。",[15,337,40],{"id":40},[11,339,340,342],{},[146,341,148],{},"。输入中文描述,它生成图片。中文 prompt 理解好是它的强项——不用翻译成英文,直接写大白话,它理解得比国外工具准。支持多种比例和风格。",[11,344,345,348],{},[146,346,347],{},"图生图","。上传一张参考图,配文字描述,生成新图。适合基于现有素材做变体,或者给草图上色细化。",[11,350,351,354],{},[146,352,353],{},"视频生成","。这是即梦区别于纯绘画工具的地方。输入文字或图片,它能生成短视频。和剪映打通,生成的视频可以直接导入剪映剪辑。不过视频生成目前质量还有限,适合做素材不适合做成品。",[11,356,357,360],{},[146,358,359],{},"智能画布","。在画布上分区生成、拼接、修改图片。类似 Photoshop 的图层概念,但用 AI 生成。适合做复杂构图。",[15,362,46],{"id":46},[11,364,365],{},"会员制：基础会员首年 ¥393（原价 ¥659\u002F年）、标准会员 ¥1099\u002F年、高级会员 ¥5199\u002F年，按积分消耗。以即梦官网为准。",[15,367,368],{"id":368},"我们的使用经验",[11,370,371],{},"我们用过即梦的免费版,主要拿它做需要中文理解的配图。几个实际感受:",[11,373,374],{},"中文 prompt 理解确实好。写\"一个穿格子衫的程序员在工位上打瞌睡,旁边放着空咖啡杯\",它出图的理解度和准确度比直接用 Midjourney 写中文强(虽然 Midjourney 本来也不擅长中文)。不用先翻译成英文,省事。",[11,376,377],{},"出图质量够用但不惊艳。和 Midjourney 比,即梦的图\"AI 味\"重一些,细节(尤其手部、文字)差一截。但在国内免费工具里,它算好的。日常配图够用,关键视觉还是得上 Midjourney。",[11,379,380],{},"免费额度大方。每天送的积分够出几十张图,日常用不用付费。这是它最大的优势——国内、免费、不用折腾网络。",[11,382,383],{},"和剪映打通是好功能。生成的图直接导入剪映做视频,工作流顺。我们做短视频的时候,即梦出图 + 剪映剪辑,不用切多个工具。",[11,385,386],{},"视频生成还在早期。试过几次文生视频,动作连贯性一般,有变形。目前适合做转场素材或者概念演示,做不了成品。",[15,388,389],{"id":389},"优缺点",[11,391,392,395],{},[146,393,394],{},"优点",":",[26,397,398,401,404,407,410],{},[29,399,400],{},"中文 prompt 理解好,不用翻译",[29,402,403],{},"国内直接用,不用科学上网",[29,405,406],{},"免费额度多,日常够用",[29,408,409],{},"和剪映打通,工作流顺",[29,411,412],{},"字节出品,稳定性和持续性有保障",[11,414,415,395],{},[146,416,417],{},"缺点",[26,419,420,423,426,429,432],{},[29,421,422],{},"出图质量不如 Midjourney,有\"AI 味\"",[29,424,425],{},"没有公开 API,没法集成到自己的系统",[29,427,428],{},"视频生成质量还在早期,做不了成品",[29,430,431],{},"中文文字渲染支持一般",[29,433,434],{},"风格偏向\"安全\",艺术感不足",[15,436,52],{"id":52},[26,438,439,442,445,448],{},[29,440,441],{},"国内用户、不想折腾网络的普通人(免费直接用)",[29,443,444],{},"做短视频、需要配图 + 剪映剪辑的创作者",[29,446,447],{},"需要中文 prompt 理解的场景",[29,449,450],{},"不适合:对出图质量有高要求的创作者(选 Midjourney);需要 API 集成的开发者;需要精确构图控制的(选 Stable Diffusion)",[15,452,63],{"id":63},[65,454,456],{"id":455},"即梦在国内能用吗","即梦在国内能用吗?",[11,458,459],{},"能,这是它最大的优势之一。字节跳动的产品,国内直接访问,不用科学上网。注册用手机号或者抖音账号,免费版直接用。",[65,461,463],{"id":462},"即梦和-midjourney-哪个好","即梦和 Midjourney 哪个好?",[11,465,466],{},"出图质量 Midjourney 强不少,但 Midjourney 要付费、要科学上网、要学 prompt。即梦免费、国内直接用、中文友好。日常配图够用选即梦,关键视觉选 Midjourney。我们两个都用,看场景。",[65,468,470],{"id":469},"即梦有免费额度吗","即梦有免费额度吗?",[11,472,473],{},"有,而且不小。每天送积分,够出几十张图。日常用不用付费。如果需要批量出图或者高清,再考虑会员。免费版是即梦的主要卖点。",[65,475,477],{"id":476},"即梦有-api-吗","即梦有 API 吗?",[11,479,480],{},"目前没有公开 API。只能通过网页或者 App 使用。如果你需要 API 集成,考虑 Stable Diffusion 或者 DALL-E 3。",{"title":80,"searchDepth":81,"depth":81,"links":482},[483,484,485,486,487,488,489],{"id":17,"depth":81,"text":18},{"id":40,"depth":81,"text":40},{"id":46,"depth":81,"text":46},{"id":368,"depth":81,"text":368},{"id":389,"depth":81,"text":389},{"id":52,"depth":81,"text":52},{"id":63,"depth":81,"text":63,"children":490},[491,492,493,494],{"id":455,"depth":90,"text":456},{"id":462,"depth":90,"text":463},{"id":469,"depth":90,"text":470},{"id":476,"depth":90,"text":477},{},"即梦","https:\u002F\u002Fjimeng.jianying.com","\u002Ftools\u002Fimage\u002Fjimeng","基础会员 ¥393\u002F年（首年，原价 ¥659\u002F年）+ 标准会员 ¥1099\u002F年 + 高级会员 ¥5199\u002F年",{"description":312},"jimeng","tools\u002Fimage\u002Fjimeng","字节跳动的 AI 绘画工具,中文 prompt 理解好,免费额度多,国内直接用",[106,107,505],"国内","字节跳动","r_24MzFupcGJwnt2rPgOYbV10bjeQGLwEg_FlGfKeDk",{"id":509,"title":510,"body":511,"category":92,"description":515,"draft":93,"experience":95,"extension":94,"hasApi":93,"meta":680,"name":510,"navigation":95,"officialUrl":681,"path":682,"pricing":683,"seo":684,"slug":685,"stem":686,"tagline":687,"tags":688,"updatedAt":110,"vendor":510,"__hash__":689},"tools\u002Ftools\u002Fimage\u002Fmidjourney.md","Midjourney",{"type":8,"value":512,"toc":666},[513,516,518,522,530,534,536,541,546,552,558,560,563,565,568,571,574,577,580,582,586,600,604,620,622,636,638,642,645,649,652,656,659,663],[11,514,515],{},"Midjourney 是 Midjourney 公司(独立团队,未公开融资)推出的 AI 绘画工具,2022 年上线。它主要能做一件事:你输入文字描述,它生成图片。出图质量在同类工具里公认最高,艺术感和细节表现突出。最早只能通过 Discord 用,后来出了网页版。",[15,517,18],{"id":17},[20,519,520],{},[11,521,319],{},[26,523,524,527],{},[29,525,526],{},"Midjourney 2026 年仍以订阅制提供网页版与 Discord 入口，具体版本号和价格以官网为准。",[29,528,529],{},"当前官网为 Cloudflare 保护页面，本站未能抓取实时定价，价格请以结账页为准。",[20,531,532],{},[11,533,335],{},[15,535,40],{"id":40},[11,537,538,540],{},[146,539,148],{},"。输入一段文字描述(英文为主),它生成 4 张图供你选。核心命令是 \u002Fimagine,prompt 写法有讲究,比如加 --ar 16:9 控制比例,加 --v 6 指定模型版本。v6 模型的写实能力比之前大幅提升,以前那种\"AI 味\"重的情况改善了不少。",[11,542,543,545],{},[146,544,347],{},"。上传一张参考图,再配文字描述,它基于参考图生成新图。我们做概念设计的时候常用,给一张草图让它细化,出图质量比自己画好太多。",[11,547,548,551],{},[146,549,550],{},"风格参考(--sref)","。给一张风格图,它把这种风格套到你的 prompt 上。比纯文字描述风格更准,适合需要统一视觉风格的项目。",[11,553,554,557],{},[146,555,556],{},"局部重绘(Vary Region)","。在生成的图里框选一块区域,重新生成。改个细节不用整张重来,出图后微调很方便。",[15,559,46],{"id":46},[11,561,562],{},"订阅制，基础版 10 美元\u002F月起，版本与档位以 Midjourney 官网为准。",[15,564,368],{"id":368},[11,566,567],{},"我们订的是 Standard 版,主要用它做网站配图和概念设计图。几个实际感受:",[11,569,570],{},"出图质量确实是同类最强。同样一个 prompt,Midjourney 出来的图视觉冲击力比 DALL-E 3 和即梦强一截。尤其是艺术风格、氛围感的图,它最擅长。",[11,572,573],{},"prompt 要学。不是随便写一句就行,得知道加参数、指定风格、控制比例。我们团队用了两周才摸出自己常用的 prompt 模板。DALL-E 3 在这点上更省心,但出图质量差一截。",[11,575,576],{},"Discord 用起来别扭。消息刷得快,自己的图很快被淹没,得翻记录找。后来出了网页版好多了,但网页版只对有过一定出图量的用户开放。",[11,578,579],{},"v6 之后写实能力上来了。以前 Midjourney 出人像有股\"油画感\",v6 之后写实路线也能走,虽然精确度还差 Stable Diffusion 配合 ControlNet 的水平。",[15,581,389],{"id":389},[11,583,584,395],{},[146,585,394],{},[26,587,588,591,594,597],{},[29,589,590],{},"出图质量公认最高,艺术感和细节突出",[29,592,593],{},"v6 模型写实能力大幅提升",[29,595,596],{},"风格参考、局部重绘等功能实用",[29,598,599],{},"社区活跃,prompt 模板和教程多",[11,601,602,395],{},[146,603,417],{},[26,605,606,609,612,615,617],{},[29,607,608],{},"不支持免费试用,最低 10 美元\u002F月",[29,610,611],{},"国内需要科学上网,付费需要海外信用卡",[29,613,614],{},"prompt 有学习成本,不是写句中文就能出好图",[29,616,425],{},[29,618,619],{},"Discord 体验一般,网页版还在逐步开放",[15,621,52],{"id":52},[26,623,624,627,630,633],{},[29,625,626],{},"对出图质量有高要求的设计师、创作者",[29,628,629],{},"做概念设计、品牌视觉、营销配图的团队",[29,631,632],{},"愿意花时间学 prompt 写法的人",[29,634,635],{},"不适合:只想用中文、不想折腾网络的普通用户;需要 API 集成的开发者;预算为零的学生",[15,637,63],{"id":63},[65,639,641],{"id":640},"midjourney-在国内能用吗","Midjourney 在国内能用吗?",[11,643,644],{},"能,但需要科学上网。Midjourney 跑在 Discord 上,你得先能访问 Discord。付费需要海外信用卡,注册需要邮箱。门槛比国产工具高不少。",[65,646,648],{"id":647},"midjourney-有免费试用吗","Midjourney 有免费试用吗?",[11,650,651],{},"没有。最低 10 美元\u002F月起步,没有免费额度。想白嫖可以看别人在 Discord 公开频道的出图,但自己不能生成。",[65,653,655],{"id":654},"midjourney-和-dall-e-3-哪个好","Midjourney 和 DALL-E 3 哪个好?",[11,657,658],{},"出图质量 Midjourney 强,prompt 理解 DALL-E 3 强。Midjourney 出来的图更好看,但 prompt 要学;DALL-E 3 随便说句话就能出图,质量够用但不如 Midjourney。要质量选 Midjourney,要省事选 DALL-E 3。",[65,660,662],{"id":661},"midjourney-有-api-吗","Midjourney 有 API 吗?",[11,664,665],{},"没有公开 API。官方目前只提供 Discord 和网页版两种使用方式。如果你需要 API 集成,考虑 Stable Diffusion 或者 DALL-E 3。",{"title":80,"searchDepth":81,"depth":81,"links":667},[668,669,670,671,672,673,674],{"id":17,"depth":81,"text":18},{"id":40,"depth":81,"text":40},{"id":46,"depth":81,"text":46},{"id":368,"depth":81,"text":368},{"id":389,"depth":81,"text":389},{"id":52,"depth":81,"text":52},{"id":63,"depth":81,"text":63,"children":675},[676,677,678,679],{"id":640,"depth":90,"text":641},{"id":647,"depth":90,"text":648},{"id":654,"depth":90,"text":655},{"id":661,"depth":90,"text":662},{},"https:\u002F\u002Fwww.midjourney.com","\u002Ftools\u002Fimage\u002Fmidjourney","订阅制，基础版 10 美元\u002F月起（具体档位与价格以官网为准）",{"description":515},"midjourney","tools\u002Fimage\u002Fmidjourney","目前出图质量最高的 AI 绘画工具,艺术感和细节表现领先,通过 Discord 和网页使用",[107,108],"ZkRrCBpFG5SSIL9lytC-Y63SAE2kZxd4XcpAi2aEWb4",{"id":691,"title":692,"body":693,"category":92,"description":697,"draft":93,"experience":93,"extension":94,"hasApi":95,"meta":756,"name":757,"navigation":95,"officialUrl":758,"path":759,"pricing":760,"seo":761,"slug":762,"stem":763,"tagline":764,"tags":765,"updatedAt":110,"vendor":766,"__hash__":767},"tools\u002Ftools\u002Fimage\u002Fnano-banana.md","Nano Banana",{"type":8,"value":694,"toc":746},[695,698,700,704,715,717,720,722,725,727,733,735,739,742,744],[11,696,697],{},"Nano Banana Pro 是 Google 的图像生成模型，放在 Gemini、Google Flow 与搜索里使用，强调多图参考、文字渲染和 4K 输出。做海报、产品图和视频分镜都能用，具体可用性与额度取决于 Google AI 订阅档位。",[15,699,18],{"id":17},[20,701,702],{},[11,703,24],{},[26,705,706,709,712],{},[29,707,708],{},"Nano Banana Pro 出现在 Google Flow 与 Runway 的模型列表里，支持 2K\u002F4K 图像生成（Google AI 订阅页、Runway 定价页，2026-11 抓取）。",[29,710,711],{},"在 Gemini App、Google Flow 与 Search 中可用，具体额度随 Google AI Plus\u002FPro\u002FUltra 档位变化。",[29,713,714],{},"API 通过 Google AI Studio \u002F Vertex AI 提供，价格以官网为准。",[15,716,40],{"id":40},[11,718,719],{},"Nano Banana Pro 是 Google 的图像生成模型，放在 Gemini、Google Flow 与搜索里使用，强调多图参考、文字渲染和 4K 输出。做海报、产品图和视频分镜都能用，具体可用性与额度取决于 Google AI 订阅档位。 具体功能边界以官网文档为准，本站只列出能确认的能力，不替厂商写宣传语。",[15,721,46],{"id":46},[11,723,724],{},"含在 Google AI 订阅中（价格以官网为准）。价格和优惠随时可能调整，下单前请以官网结账页为准。",[15,726,52],{"id":52},[26,728,729,731],{},[29,730,57],{},[29,732,60],{},[15,734,63],{"id":63},[65,736,738],{"id":737},"nano-banana-pro-是免费的吗","Nano Banana Pro 是免费的吗？",[11,740,741],{},"含在 Google AI 订阅中（价格以官网为准）。免费额度和计费方式以官网为准。",[65,743,75],{"id":74},[11,745,78],{},{"title":80,"searchDepth":81,"depth":81,"links":747},[748,749,750,751,752],{"id":17,"depth":81,"text":18},{"id":40,"depth":81,"text":40},{"id":46,"depth":81,"text":46},{"id":52,"depth":81,"text":52},{"id":63,"depth":81,"text":63,"children":753},[754,755],{"id":737,"depth":90,"text":738},{"id":74,"depth":90,"text":75},{},"Nano Banana Pro","https:\u002F\u002Fgemini.google.com","\u002Ftools\u002Fimage\u002Fnano-banana","含在 Google AI 订阅中（价格以官网为准）",{"description":697},"nano-banana","tools\u002Fimage\u002Fnano-banana","Google 的图像生成模型，在 Gemini 与 Flow 中使用，支持多图参考与 4K 输出",[107,108,109],"Google","Gt_pD0SPhXt2IGetHMFJ3uFT0g9WIC3uJG-2B5O-VZ8",{"id":769,"title":770,"body":771,"category":92,"description":775,"draft":93,"experience":93,"extension":94,"hasApi":95,"meta":832,"name":770,"navigation":95,"officialUrl":833,"path":834,"pricing":298,"seo":835,"slug":836,"stem":837,"tagline":838,"tags":839,"updatedAt":110,"vendor":770,"__hash__":840},"tools\u002Ftools\u002Fimage\u002Frecraft.md","Recraft",{"type":8,"value":772,"toc":822},[773,776,778,782,793,795,798,800,802,804,810,812,816,818,820],[11,774,775],{},"Recraft 是面向设计师的 AI 图像工具，强项是矢量图、图标、品牌风格一致性和可编辑性，不是单纯生成一张位图。适合做品牌视觉、UI 插画和电商素材。",[15,777,18],{"id":17},[20,779,780],{},[11,781,24],{},[26,783,784,787,790],{},[29,785,786],{},"Recraft 2026 年提供 V2\u002FV3\u002FV4\u002FV4.1 多代模型，支持图像、矢量、图标、背景移除与批量任务。",[29,788,789],{},"官网有 Basic、Pro、Teams 等订阅档，API 按 API Units 单独计费，具体价格以官网为准。",[29,791,792],{},"免费版生成的图片归 Recraft 所有且不可商用，付费版才有完整商用权。",[15,794,40],{"id":40},[11,796,797],{},"Recraft 是面向设计师的 AI 图像工具，强项是矢量图、图标、品牌风格一致性和可编辑性，不是单纯生成一张位图。适合做品牌视觉、UI 插画和电商素材。 具体功能边界以官网文档为准，本站只列出能确认的能力，不替厂商写宣传语。",[15,799,46],{"id":46},[11,801,263],{},[15,803,52],{"id":52},[26,805,806,808],{},[29,807,57],{},[29,809,60],{},[15,811,63],{"id":63},[65,813,815],{"id":814},"recraft-是免费的吗","Recraft 是免费的吗？",[11,817,280],{},[65,819,75],{"id":74},[11,821,78],{},{"title":80,"searchDepth":81,"depth":81,"links":823},[824,825,826,827,828],{"id":17,"depth":81,"text":18},{"id":40,"depth":81,"text":40},{"id":46,"depth":81,"text":46},{"id":52,"depth":81,"text":52},{"id":63,"depth":81,"text":63,"children":829},[830,831],{"id":814,"depth":90,"text":815},{"id":74,"depth":90,"text":75},{},"https:\u002F\u002Fwww.recraft.ai","\u002Ftools\u002Fimage\u002Frecraft",{"description":775},"recraft","tools\u002Fimage\u002Frecraft","面向设计师的 AI 图像与矢量生成工具，支持品牌风格、图标和矢量图",[106,107,108,109],"fhSJzdVVKg-jFb9XCZiwDb7k13ayS_PESDDGOq2bZr4",{"id":842,"title":843,"body":844,"category":92,"description":848,"draft":93,"experience":93,"extension":94,"hasApi":95,"meta":907,"name":908,"navigation":95,"officialUrl":497,"path":909,"pricing":910,"seo":911,"slug":912,"stem":913,"tagline":914,"tags":915,"updatedAt":110,"vendor":506,"__hash__":916},"tools\u002Ftools\u002Fimage\u002Fseedream.md","Seedream",{"type":8,"value":845,"toc":897},[846,849,851,855,866,868,871,873,876,878,884,886,890,893,895],[11,847,848],{},"Seedream 5 是字节跳动的图像生成模型，主要在即梦、豆包等字节系产品里使用，也通过 API 提供给第三方平台。中文提示词理解和亚洲人像生成是它的常见优势，适合做封面、海报和电商图。",[15,850,18],{"id":17},[20,852,853],{},[11,854,24],{},[26,856,857,860,863],{},[29,858,859],{},"Seedream 5 出现在 Pika 的图像模型列表中（Pika 定价页，2026-11 抓取），说明已对外开放给第三方平台。",[29,861,862],{},"字节系产品（即梦、豆包）使用 Seedance\u002FSeedream 系列模型，中文提示词理解是主要优势。",[29,864,865],{},"按积分或 API 调用计费，具体价格以官网为准。",[15,867,40],{"id":40},[11,869,870],{},"Seedream 5 是字节跳动的图像生成模型，主要在即梦、豆包等字节系产品里使用，也通过 API 提供给第三方平台。中文提示词理解和亚洲人像生成是它的常见优势，适合做封面、海报和电商图。 具体功能边界以官网文档为准，本站只列出能确认的能力，不替厂商写宣传语。",[15,872,46],{"id":46},[11,874,875],{},"在即梦\u002F豆包内按积分计费（价格以官网为准）。价格和优惠随时可能调整，下单前请以官网结账页为准。",[15,877,52],{"id":52},[26,879,880,882],{},[29,881,57],{},[29,883,60],{},[15,885,63],{"id":63},[65,887,889],{"id":888},"seedream-5-是免费的吗","Seedream 5 是免费的吗？",[11,891,892],{},"在即梦\u002F豆包内按积分计费（价格以官网为准）。免费额度和计费方式以官网为准。",[65,894,75],{"id":74},[11,896,78],{},{"title":80,"searchDepth":81,"depth":81,"links":898},[899,900,901,902,903],{"id":17,"depth":81,"text":18},{"id":40,"depth":81,"text":40},{"id":46,"depth":81,"text":46},{"id":52,"depth":81,"text":52},{"id":63,"depth":81,"text":63,"children":904},[905,906],{"id":888,"depth":90,"text":889},{"id":74,"depth":90,"text":75},{},"Seedream 5","\u002Ftools\u002Fimage\u002Fseedream","在即梦\u002F豆包内按积分计费（价格以官网为准）",{"description":848},"seedream","tools\u002Fimage\u002Fseedream","字节跳动的图像生成模型，即梦\u002F豆包与第三方平台均可调用，中文提示词友好",[106,107,505,109],"_EwKCALsPOfCXENoWHg8sTlUJt45Bb-nPq3nc-XmTyc",{"id":918,"title":919,"body":920,"category":92,"description":924,"draft":93,"experience":95,"extension":94,"hasApi":95,"meta":1101,"name":919,"navigation":95,"officialUrl":1102,"path":1103,"pricing":1104,"seo":1105,"slug":1106,"stem":1107,"tagline":1108,"tags":1109,"updatedAt":110,"vendor":1110,"__hash__":1111},"tools\u002Ftools\u002Fimage\u002Fstable-diffusion.md","Stable Diffusion",{"type":8,"value":921,"toc":1087},[922,925,927,931,942,946,948,954,960,966,972,974,977,979,982,985,988,991,994,997,999,1003,1020,1024,1041,1043,1057,1059,1063,1066,1070,1073,1077,1080,1084],[11,923,924],{},"Stable Diffusion 是 Stability AI 公司在 2022 年开源的 AI 绘画模型。它主要能做根据文字描述生成图像,和 Midjourney、DALL-E 3 不同的是,它是开源的——模型权重公开,可以下载到自己电脑上跑,也可以自由微调。目前 SDXL 是主流版本,SD3 也已发布但社区接受还需要时间。",[15,926,18],{"id":17},[20,928,929],{},[11,930,319],{},[26,932,933,936,939],{},[29,934,935],{},"Stability AI 2026 年官网主线为 Stable Audio、Image、Video、3D 模型与自托管授权。",[29,937,938],{},"Stable Diffusion 系列开源权重仍可本地部署和微调，具体版本以官网和 GitHub 为准。",[29,940,941],{},"商用授权与企业部署单独计费，以官网为准。",[20,943,944],{},[11,945,335],{},[15,947,40],{"id":40},[11,949,950,953],{},[146,951,952],{},"本地部署生图","。下载模型权重,装一个前端(比如 AUTOMATIC1111 或 ComfyUI),就能在自己电脑上生图。不限次数、不花钱、数据不出本地。门槛是要有一张够好的显卡,显存至少 8GB,推荐 12GB 以上。",[11,955,956,959],{},[146,957,958],{},"ControlNet 精确控制","。这是 Stable Diffusion 的杀手锏。给一张参考图,ControlNet 能提取构图、姿势、边缘线、深度图,生图时严格按这个结构走。Midjourney 和 DALL-E 3 都没有这么强的控制力。需要精确构图的人,基本只能选 Stable Diffusion。",[11,961,962,965],{},[146,963,964],{},"模型微调(LoRA)","。用少量图片训练一个 LoRA,生图时加载,就能固定某种风格或人物。比如用 20 张产品图训练,之后生图都带你的产品。开源社区有大量免费 LoRA 可用。",[11,967,968,971],{},[146,969,970],{},"云端服务","。本地显卡不够可以用云端,比如 Replicate、fal.ai,按次付费,不用自己搭环境。还有 AUTOMATIC1111 的云端部署方案。",[15,973,46],{"id":46},[11,975,976],{},"开源权重可免费本地部署；官方 API 与商用授权单独计费，以 Stability AI 官网为准。",[15,978,368],{"id":368},[11,980,981],{},"我们本地部署过,显卡是 RTX 4060 Ti 16GB,主要用来批量生成网站配图。几个实际感受:",[11,983,984],{},"学习曲线陡。装环境就折腾了一两天——Python 版本、CUDA、PyTorch、模型权重,各种依赖。ComfyUI 比 AUTOMATIC1111 难上手,但灵活度高,后期我们转向 ComfyUI。",[11,986,987],{},"ControlNet 确实强。我们要生成固定构图的图(比如人物位置、手势),用 ControlNet 的 OpenPose 和 Canny 模式,比纯文字描述准太多。这个能力是 Midjourney 给不了的。",[11,989,990],{},"批量生成是优势。本地部署不限次数,我们一次跑 100 张配图挑着用,云端按次付费扛不住这个量。",[11,992,993],{},"模型选择是门学问。CivitAI 上有几千个模型和 LoRA,挑模型花的时间不比写 prompt 少。我们最后固定用 2-3 个模型,不折腾了。",[11,995,996],{},"出图质量看模型。同一个 prompt,不同模型质量差很多。好模型能接近 Midjourney,差模型一眼就能看出 AI 味。整体来说,艺术感不如 Midjourney,但可控性强。",[15,998,389],{"id":389},[11,1000,1001,395],{},[146,1002,394],{},[26,1004,1005,1008,1011,1014,1017],{},[29,1006,1007],{},"开源免费,本地部署不限次数",[29,1009,1010],{},"ControlNet 提供精确构图控制,独此一家",[29,1012,1013],{},"模型和 LoRA 丰富,可玩性高",[29,1015,1016],{},"数据不出本地,适合敏感场景",[29,1018,1019],{},"有 API,可集成到自己的系统",[11,1021,1022,395],{},[146,1023,417],{},[26,1025,1026,1029,1032,1035,1038],{},[29,1027,1028],{},"学习曲线陡,装环境就够新手喝一壶",[29,1030,1031],{},"需要好显卡,硬件成本不低",[29,1033,1034],{},"出图质量看模型,稳定性不如商业工具",[29,1036,1037],{},"prompt 和参数调试比 Midjourney 麻烦",[29,1039,1040],{},"社区虽大但信息分散,教程质量参差不齐",[15,1042,52],{"id":52},[26,1044,1045,1048,1051,1054],{},[29,1046,1047],{},"有技术能力、愿意折腾的开发者",[29,1049,1050],{},"需要精确控制构图的设计师(ControlNet 刚需)",[29,1052,1053],{},"需要批量生图、对成本敏感的团队",[29,1055,1056],{},"不适合:不想折腾环境的普通用户;对出图质量有极高要求但不熟悉模型调优的人",[15,1058,63],{"id":63},[65,1060,1062],{"id":1061},"stable-diffusion-需要什么电脑配置","Stable Diffusion 需要什么电脑配置?",[11,1064,1065],{},"最低 8GB 显存的显卡(如 RTX 3060),推荐 12GB 以上(如 RTX 4070)。CPU 和内存影响不大,16GB 内存够用。硬盘建议留 50GB 给模型权重。Mac 也能跑但速度慢很多,不推荐。",[65,1067,1069],{"id":1068},"stable-diffusion-和-midjourney-哪个好","Stable Diffusion 和 Midjourney 哪个好?",[11,1071,1072],{},"看需求。要出图质量、要省事,选 Midjourney。要本地部署、要 ControlNet 精确控制、要批量生图,选 Stable Diffusion。我们两个都用,Midjourney 出关键图,Stable Diffusion 批量出配图。",[65,1074,1076],{"id":1075},"stable-diffusion-怎么装","Stable Diffusion 怎么装?",[11,1078,1079],{},"主流两种方式:AUTOMATIC1111(网页界面,新手友好)和 ComfyUI(节点式,灵活但难上手)。GitHub 上都有官方仓库,按 README 走。如果不想本地装,用 Replicate 或 fal.ai 的云端 API,不用装环境。",[65,1081,1083],{"id":1082},"stable-diffusion-有-api-吗","Stable Diffusion 有 API 吗?",[11,1085,1086],{},"有。Stability AI 官方提供 API,按 credit 计费。也可以用 Replicate、fal.ai 等第三方平台。如果你本地部署,可以自己包一层 API 出来,完全免费。",{"title":80,"searchDepth":81,"depth":81,"links":1088},[1089,1090,1091,1092,1093,1094,1095],{"id":17,"depth":81,"text":18},{"id":40,"depth":81,"text":40},{"id":46,"depth":81,"text":46},{"id":368,"depth":81,"text":368},{"id":389,"depth":81,"text":389},{"id":52,"depth":81,"text":52},{"id":63,"depth":81,"text":63,"children":1096},[1097,1098,1099,1100],{"id":1061,"depth":90,"text":1062},{"id":1068,"depth":90,"text":1069},{"id":1075,"depth":90,"text":1076},{"id":1082,"depth":90,"text":1083},{},"https:\u002F\u002Fstability.ai","\u002Ftools\u002Fimage\u002Fstable-diffusion","开源权重可本地部署；API 与商用授权以官网为准",{"description":924},"stable-diffusion","tools\u002Fimage\u002Fstable-diffusion","开源的 AI 绘画模型,可本地部署、自由微调,是 AI 绘画生态的基础设施",[106,107,108,109],"Stability AI","GqKhUqHAii6kGVzVU5tjvacaXf9e8QjzCHTvAncoqsU",[1113,1186,1243],{"id":1114,"title":1115,"body":1116,"category":1173,"date":1174,"description":1175,"draft":93,"extension":94,"meta":1176,"navigation":95,"path":1177,"seo":1178,"source":1142,"stem":1179,"tags":1180,"__hash__":1185},"news\u002Fnews\u002Fdaily\u002F2026-11-15.md","AI 日报 · 2026-11-15 近期要闻回顾",{"type":8,"value":1117,"toc":1168},[1118,1121,1124,1128,1145,1148,1152,1165],[11,1119,1120],{},"今日无新增权威信源发布，以下为最近 7 天内的要闻回顾（共 2 条）。完整信息请点原文链接。",[15,1122,1123],{"id":1123},"近期要闻回顾",[15,1125,1127],{"id":1126},"_1-introducing-chatgpt-wallet","1. Introducing ChatGPT Wallet",[26,1129,1130,1133,1136],{},[29,1131,1132],{},"来源：OpenAI News（原文日期 2026-11-13）",[29,1134,1135],{},"分类：产品发布",[29,1137,1138,1139],{},"原文：",[1140,1141,1142],"a",{"href":1142,"rel":1143},"https:\u002F\u002Fopenai.com\u002Findex\u002Fintroducing-chatgpt-wallet",[1144],"nofollow",[11,1146,1147],{},"Introducing ChatGPT Wallet, a secure way for agents in ChatGPT Work, Codex, and dots to complete purchases with user approval and existing card protections.",[15,1149,1151],{"id":1150},"_2-英伟达上演奥特曼20员工集体请愿让老黄回家","2. 英伟达上演“奥特曼2.0”，员工集体请愿：让老黄回家！",[26,1153,1154,1157,1159],{},[29,1155,1156],{},"来源：量子位（原文日期 2026-11-14）",[29,1158,1135],{},[29,1160,1138,1161],{},[1140,1162,1163],{"href":1163,"rel":1164},"https:\u002F\u002Fwww.qbitai.com\u002F2026\u002F11\u002F510297.html",[1144],[11,1166,1167],{},"马斯克也签名了",{"title":80,"searchDepth":81,"depth":81,"links":1169},[1170,1171,1172],{"id":1123,"depth":81,"text":1123},{"id":1126,"depth":81,"text":1127},{"id":1150,"depth":81,"text":1151},"AI 日报","2026-11-15","2026-11-15 当日无新增权威信源发布，回顾最近 7 天 2 条 AI 要闻。数据每日自动更新。",{},"\u002Fnews\u002Fdaily\u002F2026-11-15",{"title":1115,"description":1175},"news\u002Fdaily\u002F2026-11-15",[1181,1182,1183,1184],"AI日报","每日更新","权威信源","AI资讯","pK2ZeDDwRJdjym4LXFrVYODJnwR3mWF0oomb6LIO7oI",{"id":1187,"title":1188,"body":1189,"category":1173,"date":1235,"description":1236,"draft":93,"extension":94,"meta":1237,"navigation":95,"path":1238,"seo":1239,"source":1142,"stem":1240,"tags":1241,"__hash__":1242},"news\u002Fnews\u002Fdaily\u002F2026-11-14.md","AI 日报 · 2026-11-14 近期要闻回顾",{"type":8,"value":1190,"toc":1230},[1191,1193,1195,1197,1208,1210,1214,1227],[11,1192,1120],{},[15,1194,1123],{"id":1123},[15,1196,1127],{"id":1126},[26,1198,1199,1201,1203],{},[29,1200,1132],{},[29,1202,1135],{},[29,1204,1138,1205],{},[1140,1206,1142],{"href":1142,"rel":1207},[1144],[11,1209,1147],{},[15,1211,1213],{"id":1212},"_2-黄仁勋辞职英伟达33年首换帅市值蒸发超3万亿","2. 黄仁勋辞职，英伟达33年首换帅，市值蒸发超3万亿",[26,1215,1216,1219,1221],{},[29,1217,1218],{},"来源：量子位（原文日期 2026-11-13）",[29,1220,1135],{},[29,1222,1138,1223],{},[1140,1224,1225],{"href":1225,"rel":1226},"https:\u002F\u002Fwww.qbitai.com\u002F2026\u002F11\u002F509995.html",[1144],[11,1228,1229],{},"一个时代结束了",{"title":80,"searchDepth":81,"depth":81,"links":1231},[1232,1233,1234],{"id":1123,"depth":81,"text":1123},{"id":1126,"depth":81,"text":1127},{"id":1212,"depth":81,"text":1213},"2026-11-14","2026-11-14 当日无新增权威信源发布，回顾最近 7 天 2 条 AI 要闻。数据每日自动更新。",{},"\u002Fnews\u002Fdaily\u002F2026-11-14",{"title":1188,"description":1236},"news\u002Fdaily\u002F2026-11-14",[1181,1182,1183,1184],"74u2BKU3kJr6QeTNSOjutnGltojZ7johCbXt9UylzgU",{"id":1244,"title":1245,"body":1246,"category":1173,"date":1273,"description":1274,"draft":93,"extension":94,"meta":1275,"navigation":95,"path":1276,"seo":1277,"source":1265,"stem":1278,"tags":1279,"__hash__":1280},"news\u002Fnews\u002Fdaily\u002F2026-11-13.md","AI 日报 · 2026-11-13 权威信源更新",{"type":8,"value":1247,"toc":1270},[1248,1251,1255,1267],[11,1249,1250],{},"本期汇总 2026-11-13 来自 OpenAI News 的 1 条更新。内容为信源标题与摘要的客观转述，完整信息请点原文链接。",[15,1252,1254],{"id":1253},"_1-how-amplitude-ships-product-improvements-in-12-days-with-openai","1. How Amplitude ships product improvements in 1–2 days with OpenAI",[26,1256,1257,1259,1261],{},[29,1258,1132],{},[29,1260,1135],{},[29,1262,1138,1263],{},[1140,1264,1265],{"href":1265,"rel":1266},"https:\u002F\u002Fopenai.com\u002Findex\u002Famplitude",[1144],[11,1268,1269],{},"Built primarily on OpenAI models, Amplitude’s proactive product agent, Wave, identifies what to build, helps teams ship changes with coding agents, and measures the impact.",{"title":80,"searchDepth":81,"depth":81,"links":1271},[1272],{"id":1253,"depth":81,"text":1254},"2026-11-13","2026-11-13 AI 行业权威信源更新汇总，共 1 条，来源：OpenAI News。数据每日自动更新。",{},"\u002Fnews\u002Fdaily\u002F2026-11-13",{"title":1245,"description":1274},"news\u002Fdaily\u002F2026-11-13",[1181,1182,1183,1184],"kBBvbhD73coKcnRMXJH0QK3M1ogeT6cl9VEIeEnI7oM",1794695474341]