AI 模型和平台
ChatGPT 新增动画 GIF 用图像创建和编辑 2.6.
在 Google 上的首选来源中添加 Unite.AI
Verge在2026年11月4日报道，OpenAI正在ChatGPT与Images 2.6一起推出本土动画GIF创作。 更新可以将书面提示或上传的照片转换成GIF，将聊天员的图像生成工作流程扩展为可编辑动画。
生成的 GIF 的框架控制
生成后，用户可以改变动画的播放速度，修改特定帧，单独检查这些帧并分享完成的GIF. OpenAI的例子用短文本请求产生的舞猫来说明这个特征。
ChatGPT以前缺乏从请求到完成的GIF的本土单步路径。 因此，添加既改变了输出格式，也改变了修订路径：用户可以创建动画，然后在ChatGPT范围内对其时间或单个帧进行工作。 OpenAI也是称的更新使得在聊天器中更快地生成图像。
可用性和两种API模式
推出内容涵盖网络上所有ChatGPT和Codex用户以及桌面和移动应用程序。 在开发者方面，OpenAI正在添加两个API模型：GPT-Image-2.6 Flare，定位为更快生成，GPT-Image-2.6 Sunburst，定位为更高质量输出。
OpenAI 称周刊总数现已超过40亿影像，这两个总数描述了消费者和开发者产品的整体图像活动，因此它们不会隔离使用新的GIF特性。
图像背后的静态工作流程 2.6
OpenAI 称图像 2.5改进了精细细节，照明和纹理渲染，在参考图像中忠实于主体，并贯穿了连续的编辑过程，同时相对图像2.0削减了高达50%的生成延迟。 就多回合编辑而言，这意味着随着用户进行更多的编辑，更早的改动更有可能保持不变。
9月发行的版本将这些模型的改变与启动、改进和传递一个想法的控制联系起来。 用户可以与Sketch一起绘制视觉指南，从传单或产品照片的模板开始，将评论附在图像上进行定向修改，或者共享提示，以便其他人可以用不同的照片和细节进行重制。 整个桌面、移动和网络的所有ChatGPT、ChatGPT Work和Codex用户都可以使用图像2.5。
2.5 API 排队已经将吞吐量与更细的创意控制分离。 OpenAI称为GPT-Image-2.5 亮出大多数应用的默认，包括快速原型化和高容量生成，而GPT-Image-2.5 Sunburst服务于运动创意和抛光产品图像等精品工作，拥有更长的生成时间，更严格地控制编辑。
图像 2.6 新增原生动画和帧级修订以适应现有环境。 其实际后果是单ChatGPT路径从即时或相片到可编辑，可共享的GIF，而API则保留面向生成速度和视觉质量的单独的模型选项。
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
AI Models & Platforms
ChatGPT Adds Animated GIF Creation and Editing with Images 2.6
Add Unite.AI to your preferred sources on Google
The Verge reported on November 4, 2026 that OpenAI is rolling out native animated GIF creation in ChatGPT alongside Images 2.6. The update can turn a written prompt or an uploaded photo into a GIF, extending the chatbot’s image-generation workflow into editable animation.
Frame Controls for Generated GIFs
After generation, users can change how quickly the animation plays, revise specific frames, inspect those frames individually and share the finished GIF. OpenAI’s example illustrated the feature with a dancing cat generated from a short text request.
ChatGPT previously lacked a native one-step route from a request to a finished GIF. The addition therefore changes both the output format and the revision path: users can create the animation and then work on its timing or individual frames within ChatGPT. OpenAI also says the update enables faster image generation in the chatbot.
Availability and the Two API Models
The rollout covers all ChatGPT and Codex users on the web as well as desktop and mobile apps. On the developer side, OpenAI is adding two API models: GPT-Image-2.6 Flare, positioned for faster generation, and GPT-Image-2.6 Sunburst, positioned for higher-quality output.
OpenAI says the weekly total now exceeds 4 billion images across ChatGPT and the API; its September 8, 2026 Images 2.5 announcement recorded more than 3 billion weekly images across ChatGPT Images and the API’s GPT-Image models. The two totals describe overall image activity across consumer and developer products, so they do not isolate use of the new GIF feature.
The Still-Image Workflow Behind Images 2.6
OpenAI said Images 2.5 improved fine detail, lighting and texture rendering, fidelity to subjects in reference images and consistency across successive edits, while cutting generation latency by as much as 50% relative to Images 2.0. For multi-turn editing, that meant earlier changes were more likely to remain intact as users made additional edits.
The September release paired those model changes with controls for starting, refining and passing along an idea. A user could draw a visual guide with Sketch, begin from a template for a flyer or product photo, attach comments to the image for a targeted revision, or share the prompt so someone else could rework it with different photos and details. Images 2.5 was available to all ChatGPT, ChatGPT Work and Codex users across desktop, mobile and web.
The 2.5 API lineup already separated throughput from finer creative control. OpenAI called GPT-Image-2.5 Flare the default for most applications, including rapid prototyping and high-volume generation, while GPT-Image-2.5 Sunburst served premium work such as campaign creative and polished product imagery, with longer generation times and tighter control over edits.
Images 2.6 adds native animation and frame-level revision to that existing environment. The practical consequence is a single ChatGPT path from prompt or photo to an editable, shareable GIF, while the API retains separate model options oriented toward generation speed and visual quality.
Auto-fetched and translated by AI Atlas. All rights belong to the original author.