OpenAI GPT Image 2.5

13.92 credits/call
ae_openai_gpt_image_2_5

使用 GPT Image 2.5 Sunburst 根据文字描述生成 PNG 图片,支持设置图片数量、尺寸、质量和透明背景。 When generating images, size accepts only auto or WIDTHxHEIGHT pixel format; width and height must be multiples of 16, and total pixels must be between 655360 and 8294400. Common sizes include 1024x1024, 1536x1024, and 1024x1536.

Image
13.92 credits/call

Pricing

13.92 credits/call

Parameters

Optional parameters use the platform defaults when omitted.

ParameterTypeRequiredDescriptionDefault
nintegerNo请求生成的图片数量,取值范围为 1 至 10,默认为 1。1
sizestringNo输出图片尺寸。可传 auto(自动选择),或使用 WIDTHxHEIGHT 像素格式(例如 1024x1024、1536x1024、1024x1536),不支持 16:9 这类宽高比写法。WIDTH 和 HEIGHT 均须为 16 的倍数,且 WIDTH×HEIGHT 的总像素须在 655360 至 8294400 之间。默认 1024x1024。1024x1024
promptstringYes图片生成的文本提示词。建议写清主体、场景、材质、光线、构图、风格和需要渲染的文字,最多支持 32000 个字符。—
qualityenum (auto | low | medium | high)No请求的图片质量档位,可选 auto(自动选择)、low、medium 或 high,默认为 auto。auto
backgroundenum (auto | opaque | transparent)No输出图片的背景类型,可选 auto(自动选择)、opaque(不透明)或 transparent(透明)。图片以 PNG 格式输出,默认为 auto。auto
moderationenum (auto | low)No请求的内容审核级别,可选 auto 或 low,默认为 auto。low 不代表关闭审核,此参数与图片质量 quality 相互独立。auto

Invocation example

Copy the request and explore its response example.

POST/agent-api/v1/tool/executeX-Api-Key

Submit first, then query with the returned task_id until the task completes.

Request coderequest.sh
curl --request POST \
  --url 'https://agentearth.ai/agent-api/v1/tool/execute?tool_name=ae_openai_gpt_image_2_5' \
  --header "X-Api-Key: $AGENTEARTH_API_KEY" \
  --header 'Content-Type: application/json' \
  --data-raw '{
  "params": {
    "background": "opaque",
    "moderation": "auto",
    "n": 1,
    "prompt": "竖版美食海报背景,3:4 构图,中国高校食堂餐桌,暖米色木质桌面,画面四周摆放真实诱人的食堂菜品:红烧肉盖饭、牛肉面、麻婆豆腐、小笼包、清炒时蔬、一杯豆浆,热气轻柔,清晨窗边自然光,浅景深,奶油橙与米白配色,干净高级的餐饮摄影质感。画面正中央和上方留出大面积干净空白,用于后期叠加手机界面和标题。绝对不要出现任何文字、字母、数字、logo、水印、人物或手机。",
    "quality": "high",
    "size": "1024x1536"
  }
}'
Response exampleresponse.json
JSONPartial content
{
  "description": "task accepted; poll the result via query_tool with this task_id",
  "query_tool": "ae_get_result",
  "status": "created",
  "task_id": "28e84d73-99fd-4063-b593-02f37ffe3169"
}

This example shows part of the response. Copied content is valid JSON.

Guides using this tool

See how this tool fits into real workflows

What Is MCP? A Practical Guide to Model Context Protocol for AI AgentsMCP (Model Context Protocol) lets AI agents call external tools and data through a standard interface. Learn what MCP is, how it works, and how to connect {{TOOL_COUNT}} tools to Claude Code, Cursor and more.How to Connect AgentEarth to Cursor (MCP Setup, Step by Step)A hands-on guide to connecting AgentEarth to Cursor over MCP. Add the server, paste one config, verify the connection, and call {{TOOL_COUNT}} tools from natural language inside Cursor.Anime & Illustration Generation with AI AgentsLet an AI agent generate anime-style characters and scenes from one plain-English prompt — compare multiple models, render text on posters, keep a character consistent, and batch out a full illustration set without leaving your editor.The AI Agent Tool Stack for Cross-Border E-CommerceOne agent, one key, from sourcing to intelligence to creative to ads. See how a cross-border e-commerce team runs product sourcing, competitor research, image generation and SEO through a single AI agent stack.Product Listing Optimization Workflow with AI AgentsLet an AI agent run product listing optimization as one chain — mine selling points and pain points from real reviews, pick the right keywords, rewrite the copy, and generate a hero image, all from plain English.How to Connect AgentEarth to Claude Code: {{TOOL_COUNT}} Tools in One ConfigA step-by-step guide to connecting AgentEarth to Claude Code. Add one MCP config block and your agent can call {{TOOL_COUNT}} external tools in natural language — no per-tool API wiring.