OpenAI GPT Image 2.5: AI 图像生成器
Generate images from text with GPT Image 2.5. Sharper detail and up to 50% faster than GPT Image 2. Coming soon on Muapi — no subscription.
Generate images from text with GPT Image 2.5. Sharper detail and up to 50% faster than GPT Image 2. Coming soon on Muapi — no subscription.
关于此模型
GPT Image 2.5 Text to Image is OpenAI's newest image model, announced September 8, 2026, generating high-fidelity images from detailed text prompts with sharper detail, richer textures, and up to 50% lower latency than GPT Image 2. Choose the Flare tier for fast, high-volume generation or the Sunburst tier for maximum precision on premium creative work. Muapi is bringing GPT Image 2.5 online as soon as OpenAI opens general API access.
成本分析
| 提供商 | 费用 | 备注 |
|---|---|---|
| muapiapp | 发布时待定 | Coming soon — general availability 定价 not yet published by OpenAI. |
| Fal.ai | 暂不可用 | GPT Image 2.5 is 尚不可用 on Fal.ai. |
| Replicate | 暂不可用 | GPT Image 2.5 is 尚不可用 on Replicate. |
Coming soon — general availability 定价 not yet published by OpenAI.
GPT Image 2.5 is 尚不可用 on Fal.ai.
GPT Image 2.5 is 尚不可用 on Replicate.
** 竞品价格根据相似模型架构和使用层级估算。
配置参数
| 参数 | 类型 | 描述 | 默认值 |
|---|---|---|---|
| 提示词 | string | 描述图像的文本提示词 to generate. Maximum 20,000 characters. | A girl doing livestream makeup tutorials while tiny workers inside her mirror physically repaint the reflection in real time, miniature ladders, paint buckets, glowing vanity lights, cluttered modern bedroom aesthetic, insanely detailed reflections, playful surreal realism, highly creative composition, luxury commercial quality. |
| 模型 | 枚举（2 个选项） | GPT-Image-2.5 Flare is the default, fastest tier. GPT-Image-2.5 Sunburst is a slower tier tuned for maximum edit precision on premium creative work. | flare |
| 画面比例 | 枚举（10 个选项） | 输出图像画面比例。注意：画面比例为 'auto'（或未指定）的图像只能转换为 1K；1:1 不能转换为 4K，否则任务将创建失败。 | auto |
| 分辨率 | 枚举（3 个选项） | 图像分辨率。注意：画面比例为 1:1 的图像不能转换为 4K；画面比例为 'auto'（或未指定）的图像只能转换为 1K，否则任务将创建失败。 | 2K |
| 质量 | 枚举（3 个选项） | 生成质量。'low' 和 'medium' 使用更快、更便宜的后端；'high' 使用更高保真的后端。 | high |
描述图像的文本提示词 to generate. Maximum 20,000 characters.
A girl doing livestream makeup tutorials while tiny workers inside her mirror physically repaint the reflection in real time, miniature ladders, paint buckets, glowing vanity lights, cluttered modern bedroom aesthetic, insanely detailed reflections, playful surreal realism, highly creative composition, luxury commercial quality.GPT-Image-2.5 Flare is the default, fastest tier. GPT-Image-2.5 Sunburst is a slower tier tuned for maximum edit precision on premium creative work.
flare输出图像画面比例。注意：画面比例为 'auto'（或未指定）的图像只能转换为 1K；1:1 不能转换为 4K，否则任务将创建失败。
auto图像分辨率。注意：画面比例为 1:1 的图像不能转换为 4K；画面比例为 'auto'（或未指定）的图像只能转换为 1K，否则任务将创建失败。
2K生成质量。'low' 和 'medium' 使用更快、更便宜的后端；'high' 使用更高保真的后端。
high开发者文档
Write your prompt — describe the subject, style, lighting, and composition. GPT Image 2.5 follows complex, multi-part instructions more reliably than GPT Image 2.
Pick a model tier — flare (default) for speed and volume, or sunburst for maximum edit precision on premium work.
Set aspect_ratio, resolution, and quality — once live, call the endpoint below and poll for the result:
curl -X POST https://api.muapi.ai/api/v1/gpt-image-2.5-text-to-image \
-H "x-api-key: YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"prompt": "A hyperrealistic portrait of an astronaut on Martian dunes at golden hour", "model": "flare", "aspect_ratio": "16:9", "resolution": "2K"}'
request_id against /api/v1/predictions/{id}/result until status is completed.常见问答
GPT Image 2.5 is OpenAI's newest image model, announced September 8, 2026, powering ChatGPT Images 2.5 and two new API models: GPT-Image-2.5 Flare and GPT-Image-2.5 Sunburst.
Not yet. This endpoint is in early access with OpenAI. Muapi will enable it as soon as general API access opens — no action needed on your part.
Flare is the default, fastest tier with the same quality and editing gains as 2.5. Sunburst is a slower tier tuned for maximum precision on premium creative and editing workflows.
GPT Image 2.5 has better reference-photo fidelity, more precise single-element editing, more reliable multi-turn consistency, and up to 50% lower generation latency than GPT Image 2.
Yes. [GPT Image 2](/playground/gpt-image-2-text-to-image) text-to-image and image editing are live now while GPT Image 2.5 is in development.
Create a free Muapi account and API key today — the gpt-image-2.5-text-to-image endpoint will activate automatically for existing users once OpenAI opens general availability.