diff --git a/ai_service/llm/__init__.py b/ai_service/llm/__init__.py
new file mode 100644
index 0000000..059fdd8
--- /dev/null
+++ b/ai_service/llm/__init__.py
@@ -0,0 +1,50 @@
+import os
+from enum import Enum
+
+from pydantic import SecretStr
+
+
+class ProviderEnum(str, Enum):
+    """支持的 LLM 服务商"""
+    DEEPSEEK = "deepseek"
+    OPENAI = "openai"
+    TONGYI = "tongyi"
+
+class LLMFactory(object):
+
+    @staticmethod
+    def get_llm(provider: ProviderEnum, model: str = None, **kwargs):
+        if provider == ProviderEnum.DEEPSEEK:
+            from langchain_deepseek import ChatDeepSeek
+            api_key = os.getenv("DEEPSEEK_API_KEY")
+            model = model or "deepseek-chat"
+            return ChatDeepSeek(
+                api_key=SecretStr(api_key),
+                model=model,
+                streaming=True,
+                **kwargs
+            )
+
+        elif provider == ProviderEnum.OPENAI:
+            from langchain_openai import ChatOpenAI
+            api_key = os.getenv("OPENAI_API_KEY")
+            model = model or "gpt-3.5-turbo"
+            return ChatOpenAI(
+                api_key=SecretStr(api_key),
+                model=model,
+                streaming=True,
+                **kwargs
+            )
+
+        elif provider == ProviderEnum.TONGYI:
+            from langchain_community.llms import Tongyi
+            api_key = os.getenv("DASHSCOPE_API_KEY")
+            model = model or "qwen-turbo"
+            return Tongyi(
+                api_key=SecretStr(api_key),
+                model=model,
+                streaming=True,
+                **kwargs
+            )
+        else:
+            raise ValueError(f"不支持的 LLM 服务商: {provider}")
diff --git a/ai_service/llm/adapter/__init__.py b/ai_service/llm/adapter/__init__.py
new file mode 100644
index 0000000..e69de29
diff --git a/ai_service/llm/adapter/deepseek.py b/ai_service/llm/adapter/deepseek.py
new file mode 100644
index 0000000..615923e
--- /dev/null
+++ b/ai_service/llm/adapter/deepseek.py
@@ -0,0 +1,17 @@
+from langchain_deepseek import ChatDeepSeek
+
+from llm.base import MultiModalAICapability
+
+
+class DeepSeekAdapter(MultiModalAICapability):
+    def __init__(self, api_key, model, **kwargs):
+
+        self.llm = ChatDeepSeek(api_key=api_key, model=model, streaming=True)
+
+    async def chat(self, messages, **kwargs):
+        # 兼容 DeepSeek 的调用方式
+        return await self.llm.ainvoke(messages)
+
+    async def stream_chat(self, messages, **kwargs):
+        async for chunk in self.llm.astream(messages):
+            yield chunk
\ No newline at end of file
diff --git a/ai_service/llm/adapter/genai.py b/ai_service/llm/adapter/genai.py
new file mode 100644
index 0000000..f16c710
--- /dev/null
+++ b/ai_service/llm/adapter/genai.py
@@ -0,0 +1,21 @@
+# 假设有 google genai sdk
+# from google_genai import GenAI
+from llm.base import MultiModalAICapability
+
+
+class GoogleGenAIAdapter(MultiModalAICapability):
+    def __init__(self, api_key, model, **kwargs):
+        self.api_key = api_key
+        self.model = model
+        # self.llm = GenAI(api_key=api_key, model=model)
+
+    async def chat(self, messages, **kwargs):
+        # return await self.llm.chat(messages)
+        raise NotImplementedError("Google GenAI chat未实现")
+
+    async def stream_chat(self, messages, **kwargs):
+        # async for chunk in self.llm.stream_chat(messages):
+        #     yield chunk
+        raise NotImplementedError("Google GenAI stream_chat未实现")
+
+    # 其他能力同理
\ No newline at end of file
diff --git a/ai_service/llm/adapter/openai.py b/ai_service/llm/adapter/openai.py
new file mode 100644
index 0000000..1a3adb9
--- /dev/null
+++ b/ai_service/llm/adapter/openai.py
@@ -0,0 +1,25 @@
+from llm.base import MultiModalAICapability
+from langchain_openai import ChatOpenAI
+# from openai import OpenAI # 如需图片/音频/视频等API
+
+class OpenAIAdapter(MultiModalAICapability):
+    def __init__(self, api_key, model, **kwargs):
+        self.llm = ChatOpenAI(api_key=api_key, model=model, streaming=True)
+        self.api_key = api_key
+
+    async def chat(self, messages, **kwargs):
+        return await self.llm.ainvoke(messages)
+
+    async def stream_chat(self, messages, **kwargs):
+        async for chunk in self.llm.astream(messages):
+            yield chunk
+
+    # 如需图片生成（DALL·E），可实现如下
+    def create_image_task(self, prompt, **kwargs):
+        # 伪代码，需用 openai.Image.create
+        # import openai
+        # response = openai.Image.create(api_key=self.api_key, prompt=prompt, ...)
+        # return response
+        raise NotImplementedError("OpenAI 图片生成请用 openai.Image.create 实现")
+
+    # 其他能力同理
\ No newline at end of file
diff --git a/ai_service/llm/adapter/tongyi.py b/ai_service/llm/adapter/tongyi.py
new file mode 100644
index 0000000..7d7af82
--- /dev/null
+++ b/ai_service/llm/adapter/tongyi.py
@@ -0,0 +1,51 @@
+from langchain_deepseek import ChatDeepSeek
+from http import HTTPStatus
+from urllib.parse import urlparse, unquote
+from pathlib import PurePosixPath
+import requests
+from dashscope import ImageSynthesis
+import os
+
+from llm.base import MultiModalAICapability
+
+
+class TongYiAdapter(MultiModalAICapability):
+    def __init__(self, api_key, model, **kwargs):
+        self.api_key = api_key
+        self.model = model
+        self.llm = ChatDeepSeek(api_key=api_key, model=model, streaming=True)
+
+    async def chat(self, messages, **kwargs):
+        # 兼容 DeepSeek 的调用方式
+        return await self.llm.ainvoke(messages)
+
+    async def stream_chat(self, messages, **kwargs):
+        async for chunk in self.llm.astream(messages):
+            yield chunk
+
+    @staticmethod
+    def create_image_task(api_key, model, prompt: str, style='<watercolor>', size='1024*1024', n=1):
+        """创建异步图片生成任务"""
+        rsp = ImageSynthesis.async_call(
+            api_key=api_key,
+            model=model,
+            prompt=prompt,
+            n=n,
+            style=style,
+            size=size
+        )
+        if rsp.status_code == HTTPStatus.OK:
+            return rsp
+        else:
+            raise Exception(f"Failed, status_code: {rsp.status_code}, code: {rsp.code}, message: {rsp.message}")
+
+    @staticmethod
+    def fetch_image_task_status(task):
+        """获取异步图片任务状态"""
+        status = ImageSynthesis.fetch(task)
+        if status.status_code == HTTPStatus.OK:
+            return status.output.task_status
+        else:
+            raise Exception(f"Failed, status_code: {status.status_code}, code: {status.code}, message: {status.message}")
+
+    
\ No newline at end of file
diff --git a/ai_service/llm/base.py b/ai_service/llm/base.py
new file mode 100644
index 0000000..8f82046
--- /dev/null
+++ b/ai_service/llm/base.py
@@ -0,0 +1,37 @@
+from abc import ABC
+
+class MultiModalAICapability(ABC):
+    # 对话能力
+    async def chat(self, messages, **kwargs):
+        raise NotImplementedError("chat not supported by this provider")
+
+    async def stream_chat(self, messages, **kwargs):
+        raise NotImplementedError("stream_chat not supported by this provider")
+
+    # 图片生成能力
+    def create_image_task(self, prompt, **kwargs):
+        raise NotImplementedError("image generation not supported by this provider")
+
+    def fetch_image_task_status(self, task):
+        raise NotImplementedError("image task status not supported by this provider")
+
+    def fetch_image_result(self, task):
+        raise NotImplementedError("image result not supported by this provider")
+
+    # 视频生成能力
+    def create_video_task(self, prompt, **kwargs):
+        raise NotImplementedError("video generation not supported by this provider")
+
+    def fetch_video_task_status(self, task):
+        raise NotImplementedError("video task status not supported by this provider")
+
+    def fetch_video_result(self, task):
+        raise NotImplementedError("video result not supported by this provider")
+
+    # 知识库能力
+    def query_knowledge(self, query, **kwargs):
+        raise NotImplementedError("knowledge query not supported by this provider")
+
+    # 语音合成能力
+    def synthesize_speech(self, text, **kwargs):
+        raise NotImplementedError("speech synthesis not supported by this provider")
\ No newline at end of file
diff --git a/ai_service/llm/factory.py b/ai_service/llm/factory.py
new file mode 100644
index 0000000..0fce39f
--- /dev/null
+++ b/ai_service/llm/factory.py
@@ -0,0 +1,34 @@
+from .adapter.deepseek import DeepSeekAdapter
+from .adapter.genai import GoogleGenAIAdapter
+from .adapter.openai import OpenAIAdapter
+from .adapter.tongyi import TongYiAdapter
+
+
+def get_adapter(provider, api_key, model, **kwargs):
+    if provider == 'deepseek':
+        return DeepSeekAdapter(api_key, model, **kwargs)
+    elif provider == 'tongyi':
+        return TongYiAdapter(api_key, model, **kwargs)
+    elif provider == 'openai':
+        return OpenAIAdapter(api_key, model, **kwargs)
+    elif provider == 'google-genai':
+        return GoogleGenAIAdapter(api_key, model, **kwargs)
+    else:
+        raise ValueError('不支持的服务商')
+
+#  使用示例
+# adapter = get_adapter('tongyi', api_key='xxx', model='wanx_v1')
+
+# 对话
+# try:
+#     result = await adapter.chat(messages)
+# except NotImplementedError:
+#     print("该服务商不支持对话能力")
+
+# # 图片生成
+# try:
+#     task = adapter.create_image_task(prompt="一只猫")
+#     status = adapter.fetch_image_task_status(task)
+#     result = adapter.fetch_image_result(task)
+# except NotImplementedError:
+#     print("该服务商不支持图片生成")
\ No newline at end of file
diff --git a/ai_service/models/ai.py b/ai_service/models/ai.py
index 5c94dd3..edd82e4 100644
--- a/ai_service/models/ai.py
+++ b/ai_service/models/ai.py
@@ -1,5 +1,5 @@
 from sqlalchemy import (
-    Column, Integer, String, Text, DateTime, Boolean, Float, ForeignKey
+    Column, Integer, String, Text, DateTime, Boolean, Float, ForeignKey, JSON
 )
 from sqlalchemy.orm import relationship, declarative_base
 
@@ -250,4 +250,31 @@ class ChatRoleTool(Base):
     __tablename__ = 'ai_chat_role_tool'
 
     chat_role_id = Column(Integer, ForeignKey('ai_chat_role.id'), primary_key=True)
-    tool_id = Column(Integer, ForeignKey('ai_tool.id'), primary_key=True)
\ No newline at end of file
+    tool_id = Column(Integer, ForeignKey('ai_tool.id'), primary_key=True)
+
+
+class Image(CoreModel):
+    __tablename__ = 'ai_image'
+
+    user_id = Column(Integer, ForeignKey('system_users.id'), nullable=True)
+    public_status = Column(Boolean, default=False, nullable=False)
+
+    platform = Column(String(64), nullable=False)
+    model = Column(String(64), nullable=False)
+
+    prompt = Column(Text(length=2000), nullable=False)
+    width = Column(Integer, nullable=False)
+    height = Column(Integer, nullable=False)
+    options = Column(JSON, nullable=True)
+
+    status = Column(String(20), nullable=False)
+    pic_url = Column(String(2048), nullable=True)
+    error_message = Column(String(1024), nullable=True)
+
+    task_id = Column(String(1024), nullable=True)
+    buttons = Column(String(2048), nullable=True)
+
+    user = relationship("DjangoUser", backref="images")
+
+    def __str__(self):
+        return f"Image #{self.id} ({self.prompt[:30]})"
\ No newline at end of file
diff --git a/backend/ai/migrations/0005_image.py b/backend/ai/migrations/0005_image.py
new file mode 100644
index 0000000..86eaa54
--- /dev/null
+++ b/backend/ai/migrations/0005_image.py
@@ -0,0 +1,142 @@
+# Generated by Django 5.2.1 on 2025-07-21 03:06
+
+import django.db.models.deletion
+from django.conf import settings
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ("ai", "0004_alter_chatconversation_model_id_and_more"),
+        migrations.swappable_dependency(settings.AUTH_USER_MODEL),
+    ]
+
+    operations = [
+        migrations.CreateModel(
+            name="Image",
+            fields=[
+                (
+                    "id",
+                    models.BigAutoField(
+                        auto_created=True,
+                        primary_key=True,
+                        serialize=False,
+                        verbose_name="ID",
+                    ),
+                ),
+                (
+                    "remark",
+                    models.CharField(
+                        blank=True,
+                        db_comment="备注",
+                        help_text="备注",
+                        max_length=256,
+                        null=True,
+                        verbose_name="备注",
+                    ),
+                ),
+                (
+                    "creator",
+                    models.CharField(
+                        blank=True,
+                        db_comment="创建人",
+                        help_text="创建人",
+                        max_length=64,
+                        null=True,
+                        verbose_name="创建人",
+                    ),
+                ),
+                (
+                    "modifier",
+                    models.CharField(
+                        blank=True,
+                        db_comment="修改人",
+                        help_text="修改人",
+                        max_length=64,
+                        null=True,
+                        verbose_name="修改人",
+                    ),
+                ),
+                (
+                    "update_time",
+                    models.DateTimeField(
+                        auto_now=True,
+                        db_comment="修改时间",
+                        help_text="修改时间",
+                        null=True,
+                        verbose_name="修改时间",
+                    ),
+                ),
+                (
+                    "create_time",
+                    models.DateTimeField(
+                        auto_now_add=True,
+                        db_comment="创建时间",
+                        help_text="创建时间",
+                        null=True,
+                        verbose_name="创建时间",
+                    ),
+                ),
+                (
+                    "is_deleted",
+                    models.BooleanField(
+                        db_comment="是否软删除",
+                        default=False,
+                        verbose_name="是否软删除",
+                    ),
+                ),
+                (
+                    "public_status",
+                    models.BooleanField(default=False, verbose_name="是否发布"),
+                ),
+                ("platform", models.CharField(max_length=64, verbose_name="平台")),
+                ("model", models.CharField(max_length=64, verbose_name="模型")),
+                ("prompt", models.TextField(max_length=2000, verbose_name="提示词")),
+                ("width", models.IntegerField(verbose_name="图片宽度")),
+                ("height", models.IntegerField(verbose_name="图片高度")),
+                ("options", models.JSONField(null=True, verbose_name="绘制参数")),
+                ("status", models.CharField(max_length=20, verbose_name="绘画状态")),
+                (
+                    "pic_url",
+                    models.URLField(
+                        max_length=2048, null=True, verbose_name="图片地址"
+                    ),
+                ),
+                (
+                    "error_message",
+                    models.CharField(
+                        max_length=1024, null=True, verbose_name="错误信息"
+                    ),
+                ),
+                (
+                    "task_id",
+                    models.CharField(
+                        max_length=1024, null=True, verbose_name="任务编号"
+                    ),
+                ),
+                (
+                    "buttons",
+                    models.CharField(
+                        max_length=2048, null=True, verbose_name="mj buttons 按钮"
+                    ),
+                ),
+                (
+                    "user",
+                    models.ForeignKey(
+                        blank=True,
+                        db_comment="用户编号",
+                        null=True,
+                        on_delete=django.db.models.deletion.SET_NULL,
+                        to=settings.AUTH_USER_MODEL,
+                        verbose_name="用户",
+                    ),
+                ),
+            ],
+            options={
+                "verbose_name": "AI 绘画表",
+                "verbose_name_plural": "AI 绘画表",
+                "db_table": "ai_image",
+            },
+        ),
+    ]
diff --git a/backend/ai/models.py b/backend/ai/models.py
index 3f4266a..53d0ea7 100644
--- a/backend/ai/models.py
+++ b/backend/ai/models.py
@@ -330,3 +330,35 @@ class ChatMessage(CoreModel):
 
     def __str__(self):
         return self.content[:30]
+
+
+class Image(CoreModel):
+
+    user = models.ForeignKey(
+        settings.AUTH_USER_MODEL,
+        null=True, blank=True,
+        on_delete=models.SET_NULL,
+        verbose_name="用户",
+        db_comment="用户编号"
+    )
+    public_status = models.BooleanField(default=False, verbose_name='是否发布')
+
+    platform = models.CharField(max_length=64, verbose_name='平台')
+    model = models.CharField(max_length=64, verbose_name='模型')
+
+    prompt = models.TextField(max_length=2000, verbose_name='提示词')
+    width = models.IntegerField(verbose_name='图片宽度')
+    height = models.IntegerField(verbose_name='图片高度')
+    options = models.JSONField(null=True, verbose_name='绘制参数')
+
+    status = models.CharField(max_length=20, verbose_name='绘画状态')
+    pic_url = models.URLField(max_length=2048, null=True, verbose_name='图片地址')
+    error_message = models.CharField(max_length=1024, null=True, verbose_name='错误信息')
+
+    task_id = models.CharField(max_length=1024, null=True, verbose_name='任务编号')
+    buttons = models.CharField(max_length=2048, null=True, verbose_name='mj buttons 按钮')
+
+    class Meta:
+        db_table = 'ai_image'
+        verbose_name = 'AI 绘画表'
+        verbose_name_plural = verbose_name
diff --git a/web/apps/web-antd/src/views/ai/image/index.vue b/web/apps/web-antd/src/views/ai/image/index.vue
new file mode 100644
index 0000000..1d68782
--- /dev/null
+++ b/web/apps/web-antd/src/views/ai/image/index.vue
@@ -0,0 +1,11 @@
+<script setup lang="ts">
+
+</script>
+
+<template>
+  <div>dsadsa</div>
+</template>
+
+<style scoped lang="css">
+
+</style>