基于上述架构创新，该模型在CVTG-2K（复杂视觉文本生成）和LongText-Bench（长文本渲染）榜单 获得开源第一。据了解，模型尤其擅长文字密集生成任务。CVTG-2K榜单核心考察模型在图像中同时 生成多处文字的准确性，在多区域文字生成准确率上，GLM-Image的成绩位列开源模型第一； LongText-Bench（长文本渲染）榜单则考察模型渲染长文本、多行文字的准确性，覆盖招牌、海报、 PPT、对话框等8种文字密集场景，GLM-Image的成绩位列开源模型第一。 此外，模型兼具高性价比与速度优化，在API调用模式下，生成一张图片仅需0.1元，速度优化版本也即 将更新。 日前，由智谱与华为联合研发并开源的多模态图像生成模型GLM-Image，登顶Hugging Face平台 Trending榜第一，打破长期以来国外模型在开源榜首的垄断局面。 值得一提的是，该模型基于昇腾Atlas 800T A2设备和昇思MindSpore AI框架，完成从数据到训练的全流 程，是首个在国产芯片上完成全程训练的SOTA多模态模型。 而这背后，是智谱与华为"软硬协同"的深度合作，更是国产AI全产业链突破的缩影。 采用自 ...

Core Insights - The digital human industry is transitioning from a "display" model to a "functional execution" model due to the deep integration of multimodal AI, large models, and AIGC technology [1][4] - The Chinese government has introduced regulations to oversee digital human broadcasters, requiring AI-generated content to be clearly labeled and continuously remind consumers [1][6] Group 1: Industry Growth - The explosive growth of the digital human industry is driven by technological integration, enabling real-time interaction, logical reasoning, and emotional recognition [4] - Digital humans are evolving from static displays to dynamic services, capable of 24/7 live streaming and acting as government assistants or medical consultants [4] - A beauty brand achieved daily sales exceeding 500,000 yuan with a custom digital human broadcaster, showing an 18% higher conversion rate compared to human hosts [4] Group 2: Risks and Challenges - The low entry barrier of digital human technology has made it a perfect tool for spreading false information, with significant financial implications [5] - Cases of identity fraud and malicious prompts leading to unethical or illegal content generation highlight the risks associated with digital human applications [5] - A notable incident involved an AI digital human being manipulated to list products at zero cost, resulting in substantial losses for merchants [5] Group 3: Regulatory Developments - Regulatory authorities are implementing an "AI to govern AI" approach, focusing on legal, technical, and platform aspects to establish a traceability defense [6] - New regulations define compliance boundaries for commercial applications of digital humans, requiring platforms to verify operators and continuously remind consumers about AI-generated content [6] - Major platforms are adopting self-regulatory measures, such as credit evaluation systems and strict penalties for violations, to support regulatory enforcement [6]

香港科技大学深港协同创新研究院院长杨晶磊认为，与前三次工业革命"解放双手"不同，AI革命解放的 是大脑，直接冲击了人类的认知与决策核心。如果科研工作不用AI，一定会被时代淘汰。他以自己的 研究领域举例，像材料学这样传统上依赖人工实验的学科，如今也必须通过自动化和AI进行赋能，将 科研人员从繁琐的重复工作中解放出来，转向更有见地的科学发现 。 深信科创创始人、中国科学技术大学计算机学院教授杨子江抛出了一个更具冲击力的观点："科研创新 的中心正在从学术机构向拥有海量数据（603138）和算力的企业转移 。在自动驾驶、具身智能等前沿 领域，企业往往是新技术的首发地，高校反而成为了学习者。" "AI跟前三次工业革命不一样，前三次是解放双手，这次是解放大脑。""自动驾驶的所有创新几乎都来 自于企业，高校已经不具备优势了。""AI时代的产品经理不仅需要极致的技术追求，更需要'克制'的伦 理底线。" 1月15日，一场聚焦人工智能与技术扩散的思维盛宴——"扩散致远：前沿科创论坛"在河套深港科技创 新合作区隆重举行。 进入2026年，AI应用在加速落地，无论是与硬件的结合还是Agent化，用户数据泄露风险都在加大，科 技领军企 ...

智谱则在模型架构上实现了创新突破。技术人员介绍，GLM-Image没有走国外开源模型常用的技术路 线，而是采用"自回归+扩散解码器"的混合架构——简单说，就是让模型既能"读懂复杂指令"（比如生 成带文字的科普图、商业海报），又能"画对细节"（以前AI画图常出现文字错乱，这次该模型在中文生 成准确率上做到了开源模型中的首位）。 这次"全球榜首"的含金量，也体现在资本市场的反应中。此前GLM-Image开源消息发布时，看中"国产 芯片+自主模型"组合的长期价值，智谱股价单日大涨超16%。业内人士分析，从更长远看，GLM-Image 登顶不是偶然，而是国产AI全产业链协同的必然结果。这种全链条能力，不仅能让国内中小企业以更 低成本用上顶尖AI工具，更能推动国产AI技术走向全球，改变过去"国外定标准、国内跟节奏"的产业格 局。如今，GLM-Image的开源地址已在GitHub和Hugging Face同步开放，全球开发者都能免费使用。 对普通人来说，Hugging Face或许有点陌生，但在AI圈它相当于"全球开源模型博览会"——这里聚集了 谷歌、Meta等国际巨头的开源模型，也是全球开发者、企业选择AI工具的核心参 ...

All day, I see ads. Ads when I scroll social feeds, ads when I search Google, and ads on every website I go to. You're looking at some ads on this website right now (hopefully they aren't too annoying). I've lived to tell the tale — and so have you. OpenAI announced on Friday that it will start testing ads in ChatGPT for US users on its free and Go tiers, something that had been rumored for a while. If you're a brand or advertiser, this might be exciting news, but I think most of us who are merely ChatGPT ...

Artificial intelligence is the greatest investment opportunity of our lifetime. The time to invest in groundbreaking AI is now, and this stock is a steal! AI is eating the world—and the machines behind it are ravenous. Each ChatGPT query, each model update, each robotic breakthrough consumes massive amounts of energy. In fact, AI is already pushing global power grids to the brink. Wall Street is pouring hundreds of billions into artificial intelligence—training smarter chatbots, automating industries, and b ...

Artificial intelligence is the greatest investment opportunity of our lifetime. The time to invest in groundbreaking AI is now, and this stock is a steal! AI is eating the world—and the machines behind it are ravenous. Each ChatGPT query, each model update, each robotic breakthrough consumes massive amounts of energy. In fact, AI is already pushing global power grids to the brink. Wall Street is pouring hundreds of billions into artificial intelligence—training smarter chatbots, automating industries, and b ...

Artificial intelligence is the greatest investment opportunity of our lifetime. The time to invest in groundbreaking AI is now, and this stock is a steal! AI is eating the world—and the machines behind it are ravenous. Each ChatGPT query, each model update, each robotic breakthrough consumes massive amounts of energy. In fact, AI is already pushing global power grids to the brink. Wall Street is pouring hundreds of billions into artificial intelligence—training smarter chatbots, automating industries, and b ...

Artificial intelligence is the greatest investment opportunity of our lifetime. The time to invest in groundbreaking AI is now, and this stock is a steal! AI is eating the world—and the machines behind it are ravenous. Each ChatGPT query, each model update, each robotic breakthrough consumes massive amounts of energy. In fact, AI is already pushing global power grids to the brink. Wall Street is pouring hundreds of billions into artificial intelligence—training smarter chatbots, automating industries, and b ...

Core Viewpoint - OpenAI is set to initiate advertising tests within the ChatGPT application for select users in the U.S., marking a significant shift towards monetization of the popular chatbot [1] Group 1: Advertising Strategy - The advertising initiative will target free ChatGPT users and subscribers of the new low-cost Go plan, priced at $8 per month, which is currently expanding from India to the U.S. market [1] - The higher-priced premium version of ChatGPT will remain ad-free, indicating a tiered approach to monetization [1] Group 2: Financial Strategy - This decision to explore advertising is part of OpenAI's broader strategy to diversify revenue streams ahead of a potential initial public offering (IPO) [1] - The move is also aimed at offsetting the substantial costs associated with the development and operation of AI systems [1]