内容来源：https://aibusiness.com/generative-ai/openai-updates-chatgpt-images-tool

内容总结：

近日，OpenAI正式发布其图像生成模型GPT Image 1.5，该模型在生成精度、速度与创意支持方面实现显著提升，进一步加剧了AI图像生成领域的竞争。

据悉，此次更新重点强化了模型对用户指令的理解能力，图像生成速度可达前代模型的四倍。同时，新版在细节编辑、光影一致性、构图协调及文字渲染等方面均有优化，支持用户对图像进行添加、删减、融合等多种精细操作。

OpenAI在官方介绍中特别指出，ChatGPT应用及网页端新增的“图像”标签页旨在降低创作门槛，用户无需输入文字提示即可借助预设概念快速生成图像，实现“灵感激发与创意探索的无缝衔接”。公司应用业务首席执行官菲吉·西莫表示，此次升级旨在打造更贴近“创意工作室”体验的视觉创作空间。

值得注意的是，此次图像模型升级紧随OpenAI与迪士尼的合作协议落地。此前谷歌基于Gemini 3的Nano Banana Pro工具亦获市场积极反响，显示出视觉生成技术正成为AI企业竞逐的关键赛道。目前GPT Image 1.5已面向全球多数ChatGPT用户开放，企业版功能将陆续推出。

中文翻译：

由谷歌云赞助
选择你的首个生成式AI应用场景
要开始使用生成式AI，首先应关注那些能够优化人类信息交互体验的领域。

OpenAI发布GPT Image 1.5模型，聚焦提升精准度、速度与创造力
随着OpenAI推出GPT Image 1.5模型，AI图像生成领域的霸主之争日趋激烈。该公司表示，这一模型能实现前所未有的精准创作。

此次发布距谷歌基于Gemini 3的Nano Banana Pro工具面世不足一月。后者被广泛视为重大技术进步，已获得用户热烈反响。

OpenAI此次更新已面向全球多数ChatGPT用户开放，并通过API以GPT Image 1.5形式提供，带来一系列功能升级。企业级用户需等待后续开放权限。

升级核心在于提升指令遵循能力与生成速度。OpenAI称其图像生成速度最高可达前代模型的四倍。

ChatGPT应用及网页版同步新增"图像"标签页，该功能被定位为创意发生器，旨在"激发灵感，让创意探索触手可及"。

相关升级信息通过在线公告发布，其中重点强调了精准度的提升。例如编辑功能可聚焦特定细节，确保光线、构图、外观等元素在不同输出间保持更高一致性。OpenAI表示还优化了图像编辑的多个维度，包括添加、删减、组合、融合与置换功能。文字渲染能力亦获增强，现已支持更密集、更小字号的文本呈现。

公告特别指出，应用与网页端内置的预设创意方案意味着用户无需手动输入文字指令即可操作。

OpenAI应用业务首席执行官菲吉·西莫在Substack平台发文阐释本次升级理念。她坦言，虽然用户初识ChatGPT常通过文字生成图片，但现有界面最初并非为此设计，亟需构建"专为视觉创作打造的空间"。正是这一需求最终催生了此次迭代，新版设计旨在打造"更接近创意工作室"的体验。

她承诺未来将持续推出图像功能优化，包括在应答提示时增加图片使用频率，以辅助研究分析与对比展示。

此次图像更新紧随OpenAI上周与迪士尼达成的合作协议。该协议允许OpenAI在Sora视频生成模型中使用超过200个迪士尼经典角色，凸显视觉内容在AI领域日益飙升的重要性。

英文来源：

Sponsored by Google Cloud
Choosing Your First Generative AI Use Cases
To get started with generative AI, first focus on areas that can improve human experiences with information.
OpenAI's GPT Image 1.5 focuses on boosting precision, speed and creativity.
The battle for supremacy among AI image generation models is heating up with the release of OpenAI's GPT Image 1.5, which the company said delivers more precise creations than ever before.
The launch comes just under a month after the arrival of Google's Nano Banana Pro tool based on Gemini 3, which was widely considered a major step forward and has been warmly received by users.
OpenAI's update is available now for most ChatGPT users globally and available in the API as GPT Image 1.5, bringing with it an array of updates. Business and Enterprise customers will have to wait for access.
These center around an ability to better follow a user's instructions and speed, with OpenAI stating images can be generated up to four times faster than with previous models.
Also launching is a new Images tab within the ChatGPT app and browser, which is being pitched as an idea generator to "spark inspiration and make creative exploration effortless."
The upgrades were announced in a post online, where commentary focused on the added level of precision. Edits, for example, can focus in on specific details, providing more consistency with elements such as lighting, composition and appearance across outputs. OpenAI said it also improved aspects of image editing, including adding, subtracting, combining, blending and transposing. Text rendering takes a step forward, too, with denser and smaller text now able to be accommodated.
The post highlighted how an array of pre-set ideas and concepts in the app and browser mean that no written prompt is required.
Fidji Simo, CEO of applications at OpenAI, explained the rationale behind the updates in a Substack post. While a first experience with ChatGPT often involves turning a text prompt into a picture, she acknowledged the interface was not originally designed for this and a "space built for visuals" was required. This, ultimately, led to the latest iteration, which is designed to work "more like a creative studio," she added.
She went on to promise further image-focused improvements in the future, including the increased use of pictures in answers to prompts, to aid in research and provide comparisons.
The Images update follows on from OpenAI's deal with Disney announced last week, enabling it to use more than 200 of its most famous characters on its Sora video generation, underscoring the rocketing importance of visuals in AI.