dataelement · octo-patch · Aug 12, 2026
diff --git a/src/backend/bisheng/database/data/t_gpts_tools.json b/src/backend/bisheng/database/data/t_gpts_tools.json
@@ -1,4 +1,82 @@
 [
+  {
+    "name": "MiniMax Image Generation",
+    "logo": null,
+    "desc": "Generate images from a text prompt with MiniMax",
+    "api_params": [
+      {
+        "in": "query",
+        "name": "prompt",
+        "schema": {"type": "string"},
+        "required": true,
+        "description": "Text description of the image"
+      },
+      {
+        "in": "query",
+        "name": "model",
+        "schema": {"type": "string", "enum": ["image-01", "image-01-live"], "default": "image-01"},
+        "required": false,
+        "description": "MiniMax image model"
+      },
+      {
+        "in": "query",
+        "name": "aspect_ratio",
+        "schema": {"type": "string"},
+        "required": false,
+        "description": "Generated image aspect ratio"
+      },
+      {
+        "in": "query",
+        "name": "width",
+        "schema": {"type": "integer", "minimum": 512, "maximum": 2048, "multipleOf": 8},
+        "required": false,
+        "description": "Generated image width"
+      },
+      {
+        "in": "query",
+        "name": "height",
+        "schema": {"type": "integer", "minimum": 512, "maximum": 2048, "multipleOf": 8},
+        "required": false,
+        "description": "Generated image height"
+      },
+      {
+        "in": "query",
+        "name": "response_format",
+        "schema": {"type": "string", "enum": ["url", "base64"], "default": "url"},
+        "required": false,
+        "description": "Image response format"
+      },
+      {
+        "in": "query",
+        "name": "seed",
+        "schema": {"type": "integer"},
+        "required": false,
+        "description": "Seed for deterministic generation"
+      },
+      {
+        "in": "query",
+        "name": "n",
+        "schema": {"type": "integer", "minimum": 1, "maximum": 9, "default": 1},
+        "required": false,
+        "description": "Number of images to generate"
+      },
+      {
+        "in": "query",
+        "name": "prompt_optimizer",
+        "schema": {"type": "boolean", "default": false},
+        "required": false,
+        "description": "Enable prompt optimization"
+      }
+    ],
+    "tool_key": "minimax_image_generation",
+    "type": 17,
+    "is_preset": 1,
+    "is_delete": 0,
+    "user_id": 1,
+    "create_time": "2026-08-12 00:00:00",
+    "update_time": "2026-08-12 00:00:00",
+    "id": 39
+  },
   {
     "name": "时间",
     "logo": null,
@@ -1115,4 +1193,4 @@
     "user_id": 1,
     "id": 38
   }
-]
+]
diff --git a/src/backend/bisheng/database/data/t_gpts_tools_type.json b/src/backend/bisheng/database/data/t_gpts_tools_type.json
@@ -1,4 +1,20 @@
 [
+  {
+    "id": 17,
+    "name": "MiniMax",
+    "logo": "",
+    "description": "Generate images from text prompts with MiniMax image models",
+    "server_host": "",
+    "auth_method": 0,
+    "api_key": "",
+    "auth_type": "basic",
+    "is_preset": 1,
+    "user_id": null,
+    "is_delete": 0,
+    "create_time": "2026-08-12 00:00:00",
+    "update_time": "2026-08-12 00:00:00",
+    "openapi_schema": null
+  },
   {
     "id": 1,
     "name": "时间",

diff --git a/src/backend/bisheng/utils/mask_data.py b/src/backend/bisheng/utils/mask_data.py
@@ -23,6 +23,7 @@ def __init__(self):
             'api_secret': self.mask_api_key,
             'openai_api_key': self.mask_api_key,
             'siliconflow_api_key': self.mask_api_key,
+            'minimax_api_key': self.mask_api_key,
             'email_password': self.mask_api_key,
             'jina_api_key': self.mask_api_key,
             'app_secret': self.mask_api_key,
@@ -42,6 +43,7 @@ def __init__(self):
             'api_secret': r'^\*+$',
             'openai_api_key': r'^\*+$',
             'siliconflow_api_key': r'^\*+$',
+            'minimax_api_key': r'^\*+$',
             'email_password': r'^\*+$',
             'jina_api_key': r'^\*+$',
             'app_secret': r'^\*+$',

diff --git a/src/backend/bisheng_langchain/gpts/tools/api_tools/__init__.py b/src/backend/bisheng_langchain/gpts/tools/api_tools/__init__.py
@@ -3,6 +3,7 @@
 # from .eastmoney import
 from bisheng_langchain.gpts.tools.api_tools.firecrawl import FireCrawl
 from bisheng_langchain.gpts.tools.api_tools.jina import JinaTool
+from bisheng_langchain.gpts.tools.api_tools.minimax_image import MiniMaxImage
 from bisheng_langchain.gpts.tools.api_tools.silicon_flow import SiliconFlow
 from bisheng_langchain.gpts.tools.message.dingding import DingdingMessageTool
 from bisheng_langchain.gpts.tools.message.email import EmailMessageTool
@@ -74,6 +75,9 @@
     'silicon_stable_diffusion': (SiliconFlow.get_api_tool, ['siliconflow_api_key']),
     'silicon_flux': (SiliconFlow.get_api_tool, ['siliconflow_api_key'])
 }
+_MINIMAX_IMAGE_TOOLS: Dict[str, Tuple[Callable[[KwArg(Any)], BaseTool], List[str]]] = {
+    "minimax_image_generation": (MiniMaxImage.get_api_tool, ["minimax_api_key", "minimax_base_url"])
+}
 dingding_class_methods = [
     method for method in DingdingMessageTool.__dict__
     if isinstance(DingdingMessageTool.__dict__[method], classmethod)
@@ -120,6 +124,7 @@
 ALL_API_TOOLS.update(_FIRE_TOOLS)
 ALL_API_TOOLS.update(_JINA_TOOLS)
 ALL_API_TOOLS.update(_SILICON_TOOLS)
+ALL_API_TOOLS.update(_MINIMAX_IMAGE_TOOLS)
 ALL_API_TOOLS.update(_DING_TOOLS)
 ALL_API_TOOLS.update(_EMAIL_TOOLS)
 ALL_API_TOOLS.update(_FEISHU_TOOLS)

diff --git a/src/backend/bisheng_langchain/gpts/tools/api_tools/minimax_image.py b/src/backend/bisheng_langchain/gpts/tools/api_tools/minimax_image.py
@@ -0,0 +1,79 @@
+import json
+from typing import Any, Literal
+
+import requests
+from langchain_core.tools import BaseTool
+from pydantic import BaseModel, Field, model_validator
+
+from bisheng_langchain.gpts.tools.api_tools.base import APIToolBase, MultArgsSchemaTool
+from bisheng_langchain.gpts.tools.api_tools.minimax_image_core import (
+    build_image_generation_payload,
+    parse_image_generation_response,
+)
+
+
+class MiniMaxImageInput(BaseModel):
+    prompt: str = Field(description="Text description of the image", min_length=1, max_length=1500)
+    model: Literal["image-01", "image-01-live"] = "image-01"
+    aspect_ratio: Literal["1:1", "16:9", "4:3", "3:2", "2:3", "3:4", "9:16", "21:9"] | None = None
+    width: int | None = Field(default=None, ge=512, le=2048, multiple_of=8)
+    height: int | None = Field(default=None, ge=512, le=2048, multiple_of=8)
+    response_format: Literal["url", "base64"] = "url"
+    seed: int | None = None
+    n: int = Field(default=1, ge=1, le=9)
+    prompt_optimizer: bool = False
+
+    @model_validator(mode="after")
+    def validate_dimensions(self) -> "MiniMaxImageInput":
+        if (self.width is None) != (self.height is None):
+            raise ValueError("width and height must be provided together")
+        return self
+
+
+class MiniMaxImage(APIToolBase):
+    minimax_api_key: str = Field(description="MiniMax API key")
+    minimax_base_url: str = Field(default="https://api.minimax.io", description="MiniMax API base URL")
+
+    def generate(
+        self,
+        prompt: str,
+        model: str = "image-01",
+        aspect_ratio: str | None = None,
+        width: int | None = None,
+        height: int | None = None,
+        response_format: str = "url",
+        seed: int | None = None,
+        n: int = 1,
+        prompt_optimizer: bool = False,
+    ) -> str:
+        """Generate images from a text prompt with MiniMax."""
+        payload = build_image_generation_payload(
+            prompt=prompt,
+            model=model,
+            aspect_ratio=aspect_ratio,
+            width=width,
+            height=height,
+            response_format=response_format,
+            seed=seed,
+            n=n,
+            prompt_optimizer=prompt_optimizer,
+        )
+        response = requests.post(
+            f"{self.minimax_base_url.rstrip('/')}/v1/image_generation",
+            json=payload,
+            headers={"Authorization": f"Bearer {self.minimax_api_key}", "Content-Type": "application/json"},
+            timeout=self.request_timeout,
+        )
+        response.raise_for_status()
+        result = parse_image_generation_response(response.json(), response_format)
+        return json.dumps(result, ensure_ascii=False)
+
+    @classmethod
+    def get_api_tool(cls, name: str, **kwargs: Any) -> BaseTool:
+        instance = cls(**kwargs)
+        return MultArgsSchemaTool(
+            name=name,
+            description=instance.generate.__doc__,
+            func=instance.generate,
+            args_schema=MiniMaxImageInput,
+        )
diff --git a/src/backend/bisheng_langchain/gpts/tools/api_tools/minimax_image_core.py b/src/backend/bisheng_langchain/gpts/tools/api_tools/minimax_image_core.py
@@ -0,0 +1,54 @@
+from typing import Any
+
+
+def build_image_generation_payload(
+    *,
+    prompt: str,
+    model: str,
+    aspect_ratio: str | None,
+    width: int | None,
+    height: int | None,
+    response_format: str,
+    seed: int | None,
+    n: int,
+    prompt_optimizer: bool,
+) -> dict[str, Any]:
+    payload: dict[str, Any] = {
+        "model": model,
+        "prompt": prompt,
+        "response_format": response_format,
+        "n": n,
+        "prompt_optimizer": prompt_optimizer,
+    }
+    optional_fields = {
+        "aspect_ratio": aspect_ratio,
+        "width": width,
+        "height": height,
+        "seed": seed,
+    }
+    payload.update({key: value for key, value in optional_fields.items() if value is not None})
+    return payload
+
+
+def parse_image_generation_response(payload: dict[str, Any], response_format: str) -> dict[str, Any]:
+    base_response = payload.get("base_resp")
+    if not isinstance(base_response, dict) or base_response.get("status_code") != 0:
+        message = base_response.get("status_msg") if isinstance(base_response, dict) else None
+        raise ValueError(message or "MiniMax image generation failed")
+
+    data = payload.get("data")
+    if not isinstance(data, dict):
+        raise ValueError("MiniMax image generation returned no image data")
+    response_key = "image_urls" if response_format == "url" else "image_base64"
+    images = data.get(response_key)
+    if not isinstance(images, list) or not images or not all(isinstance(image, str) for image in images):
+        raise ValueError("MiniMax image generation returned invalid image data")
+
+    metadata = payload.get("metadata")
+    metadata = metadata if isinstance(metadata, dict) else {}
+    return {
+        "response_format": response_format,
+        "images": images,
+        "success_count": metadata.get("success_count", len(images)),
+        "failed_count": metadata.get("failed_count", 0),
+    }
diff --git a/src/backend/test/tool/test_minimax_image.py b/src/backend/test/tool/test_minimax_image.py
@@ -0,0 +1,76 @@
+import importlib.util
+import unittest
+from pathlib import Path
+
+MODULE_PATH = Path(__file__).parents[2] / "bisheng_langchain" / "gpts" / "tools" / "api_tools" / "minimax_image_core.py"
+SPEC = importlib.util.spec_from_file_location("minimax_image_core", MODULE_PATH)
+assert SPEC and SPEC.loader
+MODULE = importlib.util.module_from_spec(SPEC)
+SPEC.loader.exec_module(MODULE)
+build_image_generation_payload = MODULE.build_image_generation_payload
+parse_image_generation_response = MODULE.parse_image_generation_response
+
+
+class TestMiniMaxImagePayload(unittest.TestCase):
+    def test_builds_supported_request_fields(self):
+        payload = build_image_generation_payload(
+            prompt="A small robot reading a book",
+            model="image-01-live",
+            aspect_ratio=None,
+            width=1024,
+            height=768,
+            response_format="base64",
+            seed=7,
+            n=2,
+            prompt_optimizer=True,
+        )
+
+        self.assertEqual(
+            payload,
+            {
+                "model": "image-01-live",
+                "prompt": "A small robot reading a book",
+                "width": 1024,
+                "height": 768,
+                "response_format": "base64",
+                "seed": 7,
+                "n": 2,
+                "prompt_optimizer": True,
+            },
+        )
+
+    def test_parses_url_response_and_metadata(self):
+        result = parse_image_generation_response(
+            {
+                "data": {"image_urls": ["https://example.test/image.png"]},
+                "metadata": {"success_count": 1, "failed_count": 0},
+                "base_resp": {"status_code": 0, "status_msg": "success"},
+            },
+            "url",
+        )
+
+        self.assertEqual(result["images"], ["https://example.test/image.png"])
+        self.assertEqual(result["success_count"], 1)
+        self.assertEqual(result["failed_count"], 0)
+
+    def test_parses_base64_response(self):
+        result = parse_image_generation_response(
+            {
+                "data": {"image_base64": ["aW1hZ2U="]},
+                "base_resp": {"status_code": 0, "status_msg": "success"},
+            },
+            "base64",
+        )
+
+        self.assertEqual(result["images"], ["aW1hZ2U="])
+
+    def test_rejects_provider_error(self):
+        with self.assertRaisesRegex(ValueError, "invalid request"):
+            parse_image_generation_response(
+                {"base_resp": {"status_code": 2013, "status_msg": "invalid request"}},
+                "url",
+            )
+
+
+if __name__ == "__main__":
+    unittest.main()
diff --git a/src/frontend/platform/public/locales/en-US/tool.json b/src/frontend/platform/public/locales/en-US/tool.json
@@ -387,6 +387,8 @@
     "fieldCannotBeEmpty": "{{label}} cannot be empty",
     "jinaApiKeyLabel": "Jina API Key",
     "siliconflowApiKeyLabel": "SiliconFlow API Key",
+    "minimaxApiKeyLabel": "MiniMax API Key",
+    "minimaxApiBaseUrlLabel": "MiniMax API Base URL",
     "uploadFile": "Upload File",
     "organizeKnowledgeBase": "Organize Knowledge Base",
     "个人知识库": "Personal Knowledge Base",
@@ -406,4 +408,4 @@
     "写入文件内容desc": "Append text to a file or create one if it does not exist.",
     "替换文件指定行范围内容": "Replace File Line Range",
     "替换文件指定行范围内容desc": "Replace a specified range of lines in a file."
-}
+}
diff --git a/src/frontend/platform/public/locales/ja/tool.json b/src/frontend/platform/public/locales/ja/tool.json
@@ -387,6 +387,8 @@
     "fieldCannotBeEmpty": "{{label}}は空にできません",
     "jinaApiKeyLabel": "Jina API Key",
     "siliconflowApiKeyLabel": "SiliconFlow API Key",
+    "minimaxApiKeyLabel": "MiniMax API Key",
+    "minimaxApiBaseUrlLabel": "MiniMax API Base URL",
     "uploadFile": "ファイルをアップロード",
     "organizeKnowledgeBase": "ナレッジベース整理",
     "个人知识库": "個人ナレッジベース",
@@ -406,4 +408,4 @@
     "写入文件内容desc": "テキストを追記し、存在しない場合は新規作成します。",
     "替换文件指定行范围内容": "行範囲を置換",
     "替换文件指定行范围内容desc": "ファイル内の指定行範囲を置き換えます。"
-}
+}