跳转到内容

GPT Image 2.5 Text to Image

集成

模型 id openai/gpt-image-2.5/flare/text-to-image
# Run the model; wait up to 20 s for the finished result in the same call
curl -X POST "https://api.doitong.com/v1/run/openai/gpt-image-2.5/flare/text-to-image?wait=20" \
  -H "Authorization: Bearer $DOITONG_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "A handwritten recipe card titled \"Apple Pie\" on a wooden kitchen table, neat cursive ingredient list, a slice of pie on a plate beside it, cinnamon sticks, warm afternoon light, photorealistic.",
    "quality": "medium"
  }'

# Still "queued" or "running"? Poll with the id from the response:
curl "https://api.doitong.com/v1/requests/REQUEST_ID?wait=20" \
  -H "Authorization: Bearer $DOITONG_API_KEY"

此请求: $0.013 13 API 积分 在此页面运行

获取 API 密钥
OpenAI 文生图 #typography#transparent#instruction-following

OpenAI 的图像模型,指令遵循能力强,文字渲染清晰,并支持透明背景。

示例

输出 · ~24 秒
A handwritten recipe card titled "Apple Pie" on a wooden kitchen table, neat cursive ingredient list, a slice of pie ...

提示词A handwritten recipe card titled "Apple Pie" on a wooden kitchen table, neat cursive ingredient list, a slice of pie on a plate beside it, cinnamon sticks, warm afternoon light, photorealistic.

medium$0.013 · 13 API 积分

构建请求

控件根据模型架构生成。更改控件时,代码和价格会同步更新。

•必填 193 / 32000

用于生成图像的提示词。

image_size 默认 landscape_4_3

输出尺寸接受预设名称、{width, height} 对象,或让模型选择最佳尺寸的 'auto'。所有具体尺寸都要求各边可被 16 整除、边长不超过 3840px、宽高比不超过 3:1,且总像素数为 655,360–8,294,400。

background 默认 auto

结果图像的背景。

quality

图像质量选择;'auto' 将决定权交给模型。更高设置会带来更多细节,但延迟更长且 token 用量更高。

num_images 默认 1

要生成的图像数量。

高级 output_format · output_compression
output_format 默认 png

输出图像的文件格式。

仅当 output_format 为 'jpeg' 或 'webp' 时可用的压缩设置;允许范围为 0–100。

POST /run/openai/gpt-image-2.5/flare/text-to-image?wait=20 报价 → $0.013
# Run the model; wait up to 20 s for the finished result in the same call
curl -X POST "https://api.doitong.com/v1/run/openai/gpt-image-2.5/flare/text-to-image?wait=20" \
  -H "Authorization: Bearer $DOITONG_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "A handwritten recipe card titled \"Apple Pie\" on a wooden kitchen table, neat cursive ingredient list, a slice of pie on a plate beside it, cinnamon sticks, warm afternoon light, photorealistic.",
    "quality": "medium"
  }'

# Still "queued" or "running"? Poll with the id from the response:
curl "https://api.doitong.com/v1/requests/REQUEST_ID?wait=20" \
  -H "Authorization: Bearer $DOITONG_API_KEY"

运行此请求

你的密钥会保留在此浏览器中:仅发送到 Doitong API;只有勾选“记住”后才会保存到 localStorage。 获取 API 密钥

在 Doitong 上制作

Doitong 社区使用 GPT Image 2.5 制作的公开作品。一键复用提示词。

显示的公开作品:16
A polished studio concept art board — a single composite page that reads as a high-end pre-produc...

A polished studio concept art board — a single composite page that reads as a high-end pre-production pitch deck for an animated short. NOT a vintage pencil sketch, NOT a finished movie frame, but a polished concept page in the visual treatment described below. Layout: TOP HEADER (left): bold...

Art style: Реалистичный кулинарный блог. Финальный кадр, средний план ведущей. Она улыбается свое...

Art style: Реалистичный кулинарный блог. Финальный кадр, средний план ведущей. Она улыбается своей самой широкой и дружелюбной улыбкой, глядя в камеру. Она делает призывающий жест руками, словно приглашая к общению. Перед ней на столе по-прежнему лежат аппетитные булочки. Освещение становится...

Grungy analog photo of Daphne Blake (from Serial scooby doo) watching her own movie on a 90s CRT ...

Grungy analog photo of Daphne Blake (from Serial scooby doo) watching her own movie on a 90s CRT TV in a dimly lit bedroom. The TV clearly shows animated scene from scooby doo, with a cartoon-style Daphne in her classic clothes on screen, smiling. Daphne Blake is sitting cross-legged on the floor...

Fotografia realistica scattata dall’interno di un appartamento situato a un piano abbastanza alto...

Fotografia realistica scattata dall’interno di un appartamento situato a un piano abbastanza alto, guardando verso il basso attraverso una finestra aperta su una normale strada cittadina. In strada, una folla numerosa di manifestanti avanza compatta; alcune persone indossano passamontagna. Sul...

Art style: Disney-inspired. Золушка walking further back in the frame; Косметолог standing in the...

Art style: Disney-inspired. Золушка walking further back in the frame; Косметолог standing in the near foreground. The couch and strip of reflective floor separate them. Золушка is walking on glossy floor near the open doorway further back in the frame in the midground. Torso in three-quarter...

Art style: Реалистичный кулинарный блог. Экстремально крупный план рук ведущей, разламывающих гот...

Art style: Реалистичный кулинарный блог. Экстремально крупный план рук ведущей, разламывающих готовую, горячую булочку. Булочка имеет аппетитную золотисто-коричневую корочку, густо посыпанную поджаренным кунжутом. В месте разлома видна пористая, воздушная и очень мягкая текстура мякиша...

Art style: Реалистичный кулинарный блог. Финальный кадр, средний план ведущей. Она улыбается свое...

Art style: Реалистичный кулинарный блог. Финальный кадр, средний план ведущей. Она улыбается своей самой широкой и дружелюбной улыбкой, глядя в камеру. Она делает призывающий жест руками, словно приглашая к общению. Перед ней на столе по-прежнему лежат аппетитные булочки. Освещение становится...

Square album cover for an energetic house dance single. A dancing crowd silhouette in a night clu...

Square album cover for an energetic house dance single. A dancing crowd silhouette in a night club under warm orange and golden stage lights, flying sparks and flames rising from the dance floor, misty Scottish highland mountains faintly visible in the glowing background. Bold modern clean...

关于 GPT Image 2.5 Text to Image

OpenAI 的通用图像生成器:可遵循冗长而具体的指令,清晰地排布文字和图表,并可返回带透明背景的 PNG 或 WebP,适用于贴纸、图标和产品抠图。五个 质量级别在速度、价格和细节之间进行权衡。长边尺寸最高可达 3840 px(16 的倍数, 最多 3:1);每次请求最多十张图像。

参数

发送规范名称或原生名称之一,不要同时发送两者。 规范名称

名称类型 默认值允许值 描述
prompt必填
类型string
≤ 32,000 个字符
用于生成图像的提示词。
image_size可选
类型object
默认值landscape_4_3
square_hdsquareportrait_4_3portrait_16_9landscape_4_3landscape_16_9auto
输出尺寸接受预设名称、{width, height} 对象,或让模型选择最佳尺寸的 'auto'。所有具体尺寸都要求各边可被 16 整除、边长不超过 3840px、宽高比不超过 3:1,且总像素数为 655,360–8,294,400。
background可选
类型enum
默认值auto
autotransparentopaque
结果图像的背景。
quality可选
类型enum
默认值high
autolowmediumhighxhighmax
图像质量选择;'auto' 将决定权交给模型。更高设置会带来更多细节,但延迟更长且 token 用量更高。
num_images可选 num_outputs
类型integer
默认值1
1 – 10
要生成的图像数量。
output_format可选
类型enum
默认值png
jpegpngwebp
输出图像的文件格式。
output_compression可选
类型integer | null
0 – 100
仅当 output_format 为 'jpeg' 或 'webp' 时可用的压缩设置;允许范围为 0–100。
必填 规范名称,也可接受

输出

GET /requests/req_8k2m4q7x1z9c3v5b6n0p2r4t
{
  "id": "req_8k2m4q7x1z9c3v5b6n0p2r4t",
  "object": "request",
  "model": "openai/gpt-image-2.5/flare/text-to-image",
  "status": "succeeded",
  "created_at": "2026-10-20T12:00:00Z",
  "started_at": "2026-10-20T12:00:02Z",
  "completed_at": "2026-10-20T12:00:09Z",
  "queue_position": null,
  "cost": {
    "credits": 13,
    "usd": "0.013",
    "charged": 13,
    "refunded": 0
  },
  "input": {
    "prompt": "A flat vector sticker of a smiling cactus wearing sunglasses, thick white outline, no background",
    "image_size": "square_hd",
    "quality": "medium",
    "background": "transparent"
  },
  "output": {
    "images": [
      {
        "url": "https://content.doitong.com/dev-api/2026/10/20/req_8k2m4q7x1z9c3v5b6n0p2r4t/0-output.jpeg",
        "content_type": "image/jpeg",
        "file_name": "0-output.jpeg",
        "file_size": 1843200,
        "width": 1024,
        "height": 768
      }
    ]
  },
  "outputs": [
    {
      "type": "image",
      "url": "https://content.doitong.com/dev-api/2026/10/20/req_8k2m4q7x1z9c3v5b6n0p2r4t/0-output.jpeg",
      "content_type": "image/jpeg",
      "file_size": 1843200,
      "width": 1024,
      "height": 768,
      "stored": true
    }
  ],
  "error": null,
  "metrics": {
    "inference_seconds": 6.8
  },
  "urls": {
    "self": "https://api.doitong.com/v1/requests/req_8k2m4q7x1z9c3v5b6n0p2r4t",
    "cancel": "https://api.doitong.com/v1/requests/req_8k2m4q7x1z9c3v5b6n0p2r4t/cancel"
  }
}
你的文件,由你选择每个 outputs[].url 都是存储中的副本:可永久保存(按 GB-月付费),也可免费保存 7 天,由账户设置或 ?storage= 决定。
每个模型的结构相同图像、视频和音频均包含 type、url、content_type、width、height 和 duration。output 是模型自身的结果对象。
输出字段
images
array of object {url, content_type, file_name, file_size, width, height}模型生成的图像。

更多模型:文生图

对于 AI 代理:此页面的 Markdown 版本位于 https://doitong.com/zh/developers/models/openai/gpt-image-2.5/flare/text-to-image.md. 所有模型和指南: https://doitong.com/developers/llms.txt