推出 ChatGPT Images 2.5 推出 C h a t G P T I m a g e s 2.5 ChatGPTImages2.5
更清晰的细节、更快的生成速度、更精准的编辑，以及更出色的创作与分享工具。
每周，人们通过 ChatGPT Images 和 API 中的 GPT‑Image 模型生成超过 30 亿张图片。今天，我们扩展了 ChatGPT Images 2.5 的创作能力——这是我们全新的 SOTA 图像模型，为你的创意工作流带来更清晰的细节、更精准的编辑和更快的生成速度。
Images 2.5 能呈现更自然的光照和更丰富的纹理，能更好地保留参考照片中的主体，并且能在多轮对话中更可靠地遵循编辑指令。与 Images 2.0 相比，我们还将图像生成延迟降低了最高 50%，让你能更快地生成图像并打磨创意构思。
在 ChatGPT 中，我们推出了多项产品功能，让你对创作过程拥有更多掌控。Sketch 是一项新功能，让你可以直接在 ChatGPT 中绘制草图，作为最终图像的参考。Templates（模板）让你能更轻松地开始创作一些最热门的图像格式，例如传单和产品照片。现在，你还可以直接在图像上添加评论，以实现更有针对性的编辑；你也可以分享自己使用过的提示词，让其他人用他们自己的照片和细节来尝试你的创意。
Images 2.5 现已面向所有 ChatGPT、ChatGPT Work 和 Codex 用户开放，支持桌面端、移动端和网页端。
面向开发者，我们在 API 中推出了两款新模型。GPT‑Image‑2.5 Flare 在质量、编辑和速度方面带来同样的改进，而 GPT‑Image‑2.5 Sunburst 则为精细创意工作提供更高精度的支持，生成时间也更长。
创作时的图像保真度
最有意义的图像往往植根于真实生活中的人、地点和记忆。Images 2.5 能更好地利用参考照片，将熟悉的主题转换到新的场景、视觉风格和构图中。图像主体看起来更易辨认，光影和质感更自然，独特特征也更有可能得到保留。
对于使用 API 进行开发的团队，同样的保真度让以参考图为主导的工作流程更加可靠，因此变体能够始终锚定在原始来源上。
ChatGPT Images 2.5
输入
精准编辑
我们让进行有针对性的编辑变得更加容易，这样你最终得到的图像就能更贴合你实际的想法。Images 2.5 更擅长只编辑你要求修改的部分，同时保持其余细节不变——即使面对更复杂的主体和背景也是如此。
对于使用 Images API 的开发者来说，这使用户能够更新单个元素——例如产品、背景或一段文案——同时保留周围的主题、构图和品牌处理。
以下视频由多张图片组成，展示了该模型在遵循精确编辑指令方面能力的提升。
多轮编辑一致性
在较长的 ChatGPT 对话中，Images 2.5 能在多次编辑中更可靠地遵循特定的编辑指令。较早的更改更有可能保持一致，并且每次新编辑都建立在您已完成的工作之上，而不会随着时间推移降低图像质量。
这种一致性在生产工作流中也很重要，因为开发者需要在不重建整个素材的情况下进行有针对性的更改。
以下视频由多张图片组成，展示了该模型在多轮编辑中保持图像质量的能力的提升。
智能与风格改进
Images 2.5 能更好地理解复杂的视觉指令，并将其转化为连贯的结果。包含真实世界信息的图像内容更准确，并且该模型可以处理更复杂的布局，包括透明背景。它还能更好地反映视觉风格，因此您的图像更符合您的艺术构想。
对于开发者和企业而言，这让复杂的创意简报变得更加可靠。该模型更有可能保留所要求的视觉方向、构图和细节，而不会随着指令变得更加具体而偏离方向。这在生成一系列符合品牌调性的创意素材、保持特定层级结构的 UI 概念，或适配既定版式的演示视觉时尤为重要。
使用 Sketch 将你的想法画出来
有时，解释一个想法最清晰的方式就是把它画出来。
我们正在推出 Sketch，这是一项新功能，让你可以直接在 ChatGPT 中绘图，并将其作为视觉参考。现在，你可以快速画出房间的布局、正在构思的服装轮廓，或者只是一幅有趣的涂鸦，ChatGPT 会将你的粗略画作转化为完整的图像。添加风格描述以及任何其他你希望确保能体现想法的细节。
你不需要是专业艺术家——这只是另一种让最终图像更贴近你心中所想的方式。想亲自尝试，只需在 ChatGPT 中输入“@Sketch”即可。
优化你的提示词以获得更好的结果
有时你可能对自己想创作的内容有清晰的想法——只是需要一些帮助来开始。
我们正在为一些最受欢迎的创意格式推出模板。你不必再从空白画布开始，可以选择“海报”或“周边商品”等模板，并添加要传达的信息、设计元素或风格等细节，来个性化你的创作结果。
把你最棒的想法传递下去
很多时候，你会得到一张好到不分享都可惜的图片。
现在，当你分享一张图片时，还可以选择附上生成它的提示词。这样，其他人就能基于同一个想法继续创作，加入他们自己的图片和细节，做出一个属于自己的版本。
例如，试试这个 目前正流行的提示词，看看你在80年代会是什么样子！
API 中的 Images 2.5
开发者们每天都在使用 Images API，将图像生成工作流构建到创意、营销、零售和媒体团队所使用的产品中。
我们正在 API 中发布两款新的图像模型：
GPT‑Image‑2.5 Flare 将同样的质量、编辑和速度改进带到了 API 中，并且是大多数应用场景的默认选择，能以比 GPT‑Image‑2 低 50% 的延迟生成更高质量的图像。它非常适合从创作者和社交内容到产品体验、视觉搜索、快速图像原型设计以及大批量生成等各种应用。
GPT-Image-2.5 Sunburst 专为高端视觉工作流打造，这类工作流受益于编辑过程中更精细的控制。可用于创意与编辑工作流，例如可直接投产的营销创意素材或精致的商品图片。
以下是早期客户对新模型的评价：
“GPT-Image-2.5 Flare 最让我们印象深刻的是，它非常懂得哪些内容不该改动。你可以进行有意义的编辑，同时不丢失原图的特质、构图或视觉风格。这对创作者和团队在电影、UGC 和广告领域的实际工作方式来说极其重要。而当这种控制力与速度、质量和成本相结合时，GPT-Image-2.5 Flare 确实非常出众。”
Axultan Alimkulov，Higgsfield AI 产品负责人
“Adobe Firefly 是我们的创意 AI 工作室，为创作者提供从构思到成品全流程的灵活性与控制力，将领先的 AI 模型与 Adobe 的专业级工具汇聚于一处。我们很高兴能通过 Firefly 引入 OpenAI 最新的 GPT-Image-2.5 模型来扩展这一选择，其更快的生成速度和分辨率一致性，能让图像在每一次精修中始终保持清晰逼真。”
Matt Chotin，Adobe 产品高级总监
“GPT-Image-2.5 是令人瞩目的进步。在我们的评估中，Flare 生成高质量图像的速度是 GPT-Image-2 的两到四倍。凭借改进的透明背景生成能力，它非常适合 Manus 上的创意工作，从品牌素材到演示文稿和网站皆可胜任。”
Lucky Liao，Manus 评测团队成员
Higgsfield AI
Adobe
Manus
Higgsfield AI
Adobe
Manus
我们对安全的承诺
我们致力于将图像生成打造得既实用、有创意，又安全。ChatGPT Images 2.5 建立在现有安全机制之上，通过对提示词和图像进行检查，帮助防止有害内容的输出。我们继续使用 C2PA 元数据和不可见水印，帮助识别使用我们工具生成的图像。您可以在系统卡中了解更多关于我们的评估和方法的信息。
定价与可用性
Images 2.5 今日起面向所有层级的 ChatGPT、ChatGPT Work 和 Codex 用户推出，覆盖桌面端、移动端和网页端。
GPT‑Image‑2.5 Sunburst 和 GPT‑Image‑2.5 Flare 已在 API 中提供。定价详情请见此处。
Introducing ChatGPT Images 2.5 Introducing C h a t G P T I m a g e s 2.5 ChatGPTImages2.5
Sharper details, faster generation, more precise editing, and better tools for creating and sharing.
Every week, people create more than 3 billion images across ChatGPT Images and the GPT‑Image models in the API. Today, we’re expanding what you can create with ChatGPT Images 2.5—our new state-of-the-art image model that brings sharper details, more precise editing, and faster generation to your creative workflows.
Images 2.5 produces more natural lighting and richer textures, is better at preserving the subjects in your reference photos, and follows editing instructions more reliably across multiple turns. We’ve also reduced image generation latency by up to 50% compared with Images 2.0, so you can generate images and refine your concepts more quickly.
In ChatGPT, we’re introducing product features that give you more control over the creative process. Sketch is a new feature that lets you draw directly in ChatGPT as a reference for your final image. Templates make it easier to start creating across some of the most popular image formats, like flyers and product photos. You can also now place comments directly on images for more focused editing, and you can share prompts you’ve used to let others try your ideas with their own photos and details.
Images 2.5 is available to all ChatGPT, ChatGPT Work, and Codex users across desktop, mobile, and web.
For developers, we’re introducing two new models in the API. GPT‑Image‑2.5 Flare brings the same improvements in quality, editing, and speed, and GPT‑Image‑2.5 Sunburst offers an extra level of precision for detailed creative work with longer generation times.
Image fidelity as you create
The most meaningful images feel grounded in the real-life people, places, and memories they’re based on. Images 2.5 is better at working from reference photos to transform familiar subjects across new settings, visual styles, and compositions. Image subjects look more recognizable, lighting and textures feel more natural, and distinctive features are more likely to carry through.
For teams building with the API, this same fidelity makes reference-led workflows more reliable, so variations can stay anchored to the original source.
ChatGPT Images 2.5
Input
Precision editing
We’ve made it easier for you to make focused edits, so you end up with final images that are more aligned with what you actually had in mind. Images 2.5 is better at editing only what you’ve asked for, while keeping the rest of the details the same—even with more complex subjects and backgrounds.
For developers using the Images API, this gives users the ability to update a single element—such as a product, background, or piece of copy—while preserving the subject, composition, and brand treatment around it.
The following videos are made up of multiple images, highlighting the model’s improved ability to follow precise editing instructions.
Multi-turn editing consistency
During longer ChatGPT conversations, Images 2.5 follows specific editing instructions more reliably across multiple edits. Earlier changes are more likely to stay consistent, and each new edit builds on the work you’ve already done without degrading image quality over time.
That consistency also matters in production workflows, where developers need to make targeted changes without rebuilding the entire asset.
The following videos are made up of multiple images, highlighting the model’s improved ability to maintain image quality over multiple turns.
Intelligence and style improvements
Images 2.5 is better at understanding complex visual instructions and translating them into coherent results. Images that include real-world information have more accurate content, and the model can handle more complex layouts including transparent backgrounds. It’s also better at reflecting visual styles, so your images are more aligned with your artistic vision.
For developers and businesses, this makes complex creative briefs more dependable. The model is more likely to retain the requested visual direction, composition, and individual details, instead of drifting as instructions become more specific. That matters when generating a series of on-brand creative assets, UI concepts that preserve a given hierarchy, or presentation visuals that fit a defined structure.
Use Sketch to draw your idea to life
Sometimes the clearest way to explain an idea is by drawing it.
We’re introducing Sketch, a new feature which lets you draw right in ChatGPT and use it as a visual guide. Now you can quickly sketch the layout of a room, the contour of an outfit you’re concepting, or just a funny doodle, and ChatGPT will turn your rough art into a complete image. Add a description of the style and any other details you want to make sure it reflects your vision.
You don’t need to be a professional artist—this is just another way to get your final image closer to what you have in mind. To try it yourself , just type “@Sketch” in ChatGPT.
Structure your prompts for better results
At times you may have a clear idea of what you want to make—you just need some help getting started.
We’re introducing templates for some of the most popular creative formats. Instead of starting with a blank canvas, you can choose a template like “Poster” or “Merch” and add details like information to convey, design elements, or styles to personalize your results.
Pass your best ideas along
Often you end up with an image that’s too good not to share.
Now when you share an image, you can also choose to include the prompt that got you there. Someone else can then run with the same idea, bringing in their own images and details to make a version that feels like their own.
For example, try out this prompt that’s currently going viral to see what you would’ve looked like in the ’80s!
Images 2.5 in the API
Developers are building image generation workflows into products used by creative, marketing, retail, and media teams every day with the Images API.
We’re releasing two new image models in the API:
GPT‑Image‑2.5 Flare brings the same improvements in quality, editing, and speed to the API and is the default choice for most applications, delivering higher-quality images than GPT‑Image‑2 at 50% lower latency. It’s well suited to everything from creator and social content to product experiences, visual search, rapid image prototyping, and high-volume generation.
GPT‑Image‑2.5 Sunburst is built for premium visual workflows that benefit from tighter control across edits. Use it for creative and editing workflows like production-ready campaign creative or polished product imagery.
Here’s what early customers have to say about the new models:
“What impressed us most about GPT‑Image‑2.5 Flare is how well it understands what not to change. You can make a meaningful edit without losing the character, composition or visual identity of the original image. That’s incredibly important for the way creators and teams actually work across film, UGC and advertising. And when you combine that level of control with the speed, quality and cost, GPT‑Image‑2.5 Flare really stands out.”
Axultan Alimkulov, Head of Product at Higgsfield AI
“Adobe Firefly, our creative AI studio, gives creators flexibility and control as they move from idea to finished content, bringing together leading AI models and Adobe's pro-grade tools in one place. We're excited to expand that choice with OpenAI's latest GPT‑Image‑2.5 models now in Firefly, bringing faster generation and resolution consistency that keep images sharp and photorealistic through every refinement.”
Matt Chotin, Senior Director, Product at Adobe
“GPT‑Image‑2.5 is an impressive step forward. In our evaluations, Flare delivers high-quality images at two to four times the speed of GPT‑Image‑2. With improved transparent-background generation, it is an excellent fit for creative work on Manus, from brand assets to presentations and websites.”
Lucky Liao, Evaluation Team at Manus
Higgsfield AI
Adobe
Manus
Higgsfield AI
Adobe
Manus
Our commitment to safety
We’re building image generation to be useful, creative, and safe. ChatGPT Images 2.5 builds on our existing safeguards, with checks on prompts and images to help prevent harmful outputs. We continue to use C2PA metadata and invisible watermarking to help identify images made with our tools. You can read more about our evaluations and approach in the system card .
Pricing and availability
Images 2.5 is rolling out today to ChatGPT, ChatGPT Work, and Codex users across all tiers on desktop, mobile, and web.
GPT‑Image‑2.5 Sunburst and GPT‑Image‑2.5 Flare are available in the API. See pricing details here.