diff --git a/.cursor/rules/testing-guide/testing-guide.mdc b/.cursor/rules/testing-guide/testing-guide.mdc
index bd63bef840..df33456d8f 100644
--- a/.cursor/rules/testing-guide/testing-guide.mdc
+++ b/.cursor/rules/testing-guide/testing-guide.mdc
@@ -214,6 +214,176 @@ describe('<UserProfile />', () => {
 **修复方法**: 更新了测试文件中的 mock 数据结构，使其与最新的 API 响应格式保持一致。具体修改了 `user.test.ts` 中的 `mockUserData` 对象结构。
 ```
 
+## 🎯 测试编写最佳实践
+
+### Mock 数据策略：追求"低成本的真实性" 📋
+
+**核心原则**: 测试数据应默认追求真实性，只有在引入"高昂的测试成本"时才进行简化。
+
+#### 什么是"高昂的测试成本"？
+
+"高成本"指的是测试中引入了外部依赖，使测试变慢、不稳定或复杂：
+
+- **文件 I/O 操作**：读写硬盘文件
+- **网络请求**：HTTP 调用、数据库连接
+- **系统调用**：获取系统时间、环境变量等
+
+#### ✅ 推荐做法：Mock 依赖，保留真实数据
+
+```typescript
+// ✅ 好的做法：Mock I/O 操作，但使用真实的文件内容格式
+describe('parseContentType', () => {
+  beforeEach(() => {
+    // Mock 文件读取操作（避免真实 I/O）
+    vi.spyOn(fs, 'readFileSync').mockImplementation((path) => {
+      // 但返回真实的文件内容格式
+      if (path.includes('.pdf')) return '%PDF-1.4\n%âãÏÓ'; // 真实 PDF 文件头
+      if (path.includes('.png')) return '\x89PNG\r\n\x1a\n'; // 真实 PNG 文件头
+      return '';
+    });
+  });
+
+  it('should detect PDF content type correctly', () => {
+    const result = parseContentType('/path/to/file.pdf');
+    expect(result).toBe('application/pdf');
+  });
+});
+
+// ❌ 过度简化：使用不真实的数据
+describe('parseContentType', () => {
+  it('should detect PDF content type correctly', () => {
+    // 这种简化数据没有测试价值
+    const result = parseContentType('fake-pdf-content');
+    expect(result).toBe('application/pdf');
+  });
+});
+```
+
+#### 🎯 真实标识符的价值
+
+```typescript
+// ✅ 使用真实的提供商标识符
+it('should parse OpenAI model list correctly', () => {
+  const result = parseModelString('openai', '+gpt-4,+gpt-3.5-turbo');
+  expect(result.add).toHaveLength(2);
+  expect(result.add[0].id).toBe('gpt-4');
+});
+
+// ❌ 使用占位符标识符（价值较低）
+it('should parse model list correctly', () => {
+  const result = parseModelString('test-provider', '+model1,+model2');
+  expect(result.add).toHaveLength(2);
+  // 这种测试对理解真实场景帮助不大
+});
+```
+
+### 错误处理测试：测试"行为"而非"文本" ⚠️
+
+**核心原则**: 测试应该验证程序在错误发生时的行为是可预测的，而不是验证易变的错误信息文本。
+
+#### ✅ 推荐的错误测试方式
+
+```typescript
+// ✅ 测试是否抛出错误
+it('should throw error when invalid input provided', () => {
+  expect(() => processInput(null)).toThrow();
+});
+
+// ✅ 测试错误类型（最推荐）
+it('should throw ValidationError for invalid data', () => {
+  expect(() => validateUser({})).toThrow(ValidationError);
+});
+
+// ✅ 测试错误属性而非消息文本
+it('should throw error with correct error code', () => {
+  expect(() => processPayment({})).toThrow(
+    expect.objectContaining({
+      code: 'INVALID_PAYMENT_DATA',
+      statusCode: 400,
+    }),
+  );
+});
+```
+
+#### ❌ 应避免的做法
+
+```typescript
+// ❌ 过度依赖具体错误信息文本
+it('should throw specific error message', () => {
+  expect(() => processUser({})).toThrow('用户数据不能为空，请检查输入参数');
+  // 这种测试很脆弱，错误文案稍有修改就会失败
+});
+```
+
+#### 🎯 例外情况：何时可以测试错误信息
+
+```typescript
+// ✅ 测试标准 API 错误（这是契约的一部分）
+it('should return proper HTTP error for API', () => {
+  expect(response.statusCode).toBe(400);
+  expect(response.error).toBe('Bad Request');
+});
+
+// ✅ 测试错误信息的关键部分（使用正则）
+it('should include field name in validation error', () => {
+  expect(() => validateField('email', '')).toThrow(/email/i);
+});
+```
+
+### 疑难解答：警惕模块污染 🚨
+
+**识别信号**: 当你的测试出现以下"灵异"现象时，优先怀疑模块污染：
+
+- 单独运行某个测试通过，但和其他测试一起运行就失败
+- 测试的执行顺序影响结果
+- Mock 设置看起来正确，但实际使用的是旧的 Mock 版本
+
+#### 典型场景：动态 Mock 同一模块
+
+```typescript
+// ❌ 容易出现模块污染的写法
+describe('ConfigService', () => {
+  it('should work in development mode', async () => {
+    vi.doMock('./config', () => ({ isDev: true }));
+    const { getSettings } = await import('./configService'); // 第一次加载
+    expect(getSettings().debugMode).toBe(true);
+  });
+
+  it('should work in production mode', async () => {
+    vi.doMock('./config', () => ({ isDev: false }));
+    const { getSettings } = await import('./configService'); // 可能使用缓存的旧版本！
+    expect(getSettings().debugMode).toBe(false); // ❌ 可能失败
+  });
+});
+
+// ✅ 使用 resetModules 解决模块污染
+describe('ConfigService', () => {
+  beforeEach(() => {
+    vi.resetModules(); // 清除模块缓存，确保每个测试都是干净的环境
+  });
+
+  it('should work in development mode', async () => {
+    vi.doMock('./config', () => ({ isDev: true }));
+    const { getSettings } = await import('./configService');
+    expect(getSettings().debugMode).toBe(true);
+  });
+
+  it('should work in production mode', async () => {
+    vi.doMock('./config', () => ({ isDev: false }));
+    const { getSettings } = await import('./configService');
+    expect(getSettings().debugMode).toBe(false); // ✅ 测试通过
+  });
+});
+```
+
+#### 🔧 排查和解决步骤
+
+1. **识别问题**: 测试失败时，首先问自己："是否有多个测试在 Mock 同一个模块？"
+2. **添加隔离**: 在 `beforeEach` 中添加 `vi.resetModules()`
+3. **验证修复**: 重新运行测试，确认问题解决
+
+**记住**: `vi.resetModules()` 是解决测试"灵异"失败的终极武器，当常规调试方法都无效时，它往往能一针见血地解决问题。
+
 ## 📂 测试文件组织
 
 ### 文件命名约定
@@ -320,4 +490,7 @@ git show HEAD -- path/to/component.ts | cat # 查看最新提交的修改
 - **修复原则**: 失败1-2次后寻求帮助，测试命名关注行为而非实现细节
 - **调试流程**: 复现 → 分析 → 假设 → 修复 → 验证 → 总结
 - **文件组织**: 优先在现有 `describe` 块中添加测试，避免创建冗余顶级块
+- **数据策略**: 默认追求真实性，只有高成本（I/O、网络等）时才简化
+- **错误测试**: 测试错误类型和行为，避免依赖具体的错误信息文本
+- **模块污染**: 测试"灵异"失败时，优先怀疑模块污染，使用 `vi.resetModules()` 解决
 - **安全要求**: Model 测试必须包含权限检查，并在双环境下验证通过
diff --git a/docs/self-hosting/environment-variables/model-provider.mdx b/docs/self-hosting/environment-variables/model-provider.mdx
index 72cab858d9..5d5592f19e 100644
--- a/docs/self-hosting/environment-variables/model-provider.mdx
+++ b/docs/self-hosting/environment-variables/model-provider.mdx
@@ -625,4 +625,29 @@ If you need to use Azure OpenAI to provide model services, you can refer to the
 - Default: `-`
 - Example: `-all,+qwq-32b,+deepseek-r1`
 
+## FAL
+
+### `ENABLED_FAL`
+
+- Type: Optional
+- Description: Enables FAL as a model provider by default. Set to `0` to disable the FAL service.
+- Default: `1`
+- Example: `0`
+
+### `FAL_API_KEY`
+
+- Type: Required
+- Description: This is the API key you applied for in the FAL service.
+- Default: -
+- Example: `fal-xxxxxx...xxxxxx`
+
+### `FAL_MODEL_LIST`
+
+- Type: Optional
+- Description: Used to control the FAL model list. Use `+` to add a model, `-` to hide a model, and `model_name=display_name` to customize the display name of a model. Separate multiple entries with commas. The definition syntax follows the same rules as other providers' model lists.
+- Default: `-`
+- Example: `-all,+fal-model-1,+fal-model-2=fal-special`
+
+The above example disables all models first, then enables `fal-model-1` and `fal-model-2` (displayed as `fal-special`).
+
 [model-list]: /docs/self-hosting/advanced/model-list
diff --git a/docs/self-hosting/environment-variables/model-provider.zh-CN.mdx b/docs/self-hosting/environment-variables/model-provider.zh-CN.mdx
index e3d491b3b1..4d5d4f9d04 100644
--- a/docs/self-hosting/environment-variables/model-provider.zh-CN.mdx
+++ b/docs/self-hosting/environment-variables/model-provider.zh-CN.mdx
@@ -624,4 +624,29 @@ LobeChat 在部署时提供了丰富的模型服务商相关的环境变量，
 - 默认值：`-`
 - 示例：`-all,+qwq-32b,+deepseek-r1`
 
+## FAL
+
+### `ENABLED_FAL`
+
+- 类型：可选
+- 描述：默认启用 FAL 作为模型供应商，当设为 0 时关闭 FAL 服务
+- 默认值：`1`
+- 示例：`0`
+
+### `FAL_API_KEY`
+
+- 类型：必选
+- 描述：这是你在 FAL 服务中申请的 API 密钥
+- 默认值：-
+- 示例：`fal-xxxxxx...xxxxxx`
+
+### `FAL_MODEL_LIST`
+
+- 类型：可选
+- 描述：用来控制 FAL 模型列表，使用 `+` 增加一个模型，使用 `-` 来隐藏一个模型，使用 `模型名=展示名` 来自定义模型的展示名，用英文逗号隔开。模型定义语法规则与其他 provider 保持一致。
+- 默认值：`-`
+- 示例：`-all,+fal-model-1,+fal-model-2=fal-special`
+
+上述示例表示先禁用所有模型，再启用 `fal-model-1` 和 `fal-model-2`（显示名为 `fal-special`）。
+
 [model-list]: /zh/docs/self-hosting/advanced/model-list
diff --git a/docs/self-hosting/faq/vercel-ai-image-timeout.mdx b/docs/self-hosting/faq/vercel-ai-image-timeout.mdx
new file mode 100644
index 0000000000..414577ca2b
--- /dev/null
+++ b/docs/self-hosting/faq/vercel-ai-image-timeout.mdx
@@ -0,0 +1,65 @@
+---
+title: Resolving AI Image Generation Timeout on Vercel
+description: >-
+  Learn how to resolve timeout issues when using AI image generation models like gpt-image-1 on Vercel by enabling Fluid Compute for extended execution time.
+
+tags:
+  - Vercel
+  - AI Image Generation
+  - Timeout
+  - Fluid Compute
+  - gpt-image-1
+---
+
+# Resolving AI Image Generation Timeout on Vercel
+
+## Problem Description
+
+When using AI image generation models (such as `gpt-image-1`) on Vercel, you may encounter timeout errors. This occurs because AI image generation typically requires more than 1 minute to complete, which exceeds Vercel's default function execution time limit.
+
+Common error symptoms include:
+
+- Function timeout errors during image generation
+- Failed image generation requests after approximately 60 seconds
+- "Function execution timed out" messages
+
+### Typical Log Symptoms
+
+In your Vercel function logs, you may see entries like this:
+
+```plaintext
+JUL 16 18:39:09.51 POST 504 /trpc/async/image.createImage
+Provider runtime map found for provider: openai
+```
+
+The key indicators are:
+
+- **Status Code**: `504` (Gateway Timeout)
+- **Endpoint**: `/trpc/async/image.createImage` or similar image generation endpoints
+- **Timing**: Usually occurs around 60 seconds after the request starts
+
+## Solution: Enable Fluid Compute
+
+For projects created before Vercel's dashboard update, you can resolve this issue by enabling Fluid Compute, which extends the maximum execution duration to 300 seconds.
+
+### Steps to Enable Fluid Compute (Legacy Vercel Dashboard)
+
+1. Go to your project dashboard on Vercel
+2. Navigate to the **Settings** tab
+3. Find the **Functions** section
+4. Enable **Fluid Compute** as shown in the screenshot below:
+
+![Enable Fluid Compute](https://hub-apac-1.lobeobjects.space/docs/2e2ff332c4b440b584efe1d7ba46aed5.png)
+
+5. After enabling, the maximum execution duration will be extended to 300 seconds by default
+
+### Important Notes
+
+- **For new projects**: Newer Vercel projects have Fluid Compute enabled by default, so this issue primarily affects legacy projects
+
+## Additional Resources
+
+For more information about Vercel's function limitations and Fluid Compute:
+
+- [Vercel Fluid Compute Documentation](https://vercel.com/docs/fluid-compute)
+- [Vercel Functions Limitations](https://vercel.com/docs/functions/limitations#max-duration)
diff --git a/docs/self-hosting/faq/vercel-ai-image-timeout.zh-CN.mdx b/docs/self-hosting/faq/vercel-ai-image-timeout.zh-CN.mdx
new file mode 100644
index 0000000000..287c97a9d8
--- /dev/null
+++ b/docs/self-hosting/faq/vercel-ai-image-timeout.zh-CN.mdx
@@ -0,0 +1,63 @@
+---
+title: 解决 Vercel 上 AI 绘画生图超时问题
+description: 了解如何通过开启 Fluid Compute 来解决在 Vercel 上使用 gpt-image-1 等 AI 绘画模型时遇到的超时问题。
+tags:
+  - Vercel
+  - AI 绘画
+  - 超时问题
+  - Fluid Compute
+  - gpt-image-1
+---
+
+# 解决 Vercel 上 AI 绘画生图超时问题
+
+## 问题描述
+
+在 Vercel 上使用 AI 绘画模型（如 `gpt-image-1`）时，您可能会遇到超时错误。这是因为 AI 绘画生成通常需要超过 1 分钟的时间，超出了 Vercel 默认的函数执行时间限制。
+
+常见的错误症状包括：
+
+- 图像生成过程中出现函数超时错误
+- 图像生成请求在大约 60 秒后失败
+- 出现 "函数执行超时" 的错误消息
+
+### 典型的日志现象
+
+在您的 Vercel 函数日志中，您可能会看到类似这样的条目：
+
+```plaintext
+JUL 16 18:39:09.51 POST 504 /trpc/async/image.createImage
+Provider runtime map found for provider: openai
+```
+
+关键指标包括：
+
+- **状态码**: `504`（网关超时）
+- **端点**: `/trpc/async/image.createImage` 或类似的图像生成端点
+- **时间**: 通常在请求开始后约 60 秒出现
+
+## 解决方案：开启 Fluid Compute
+
+对于在 Vercel 控制台更新前创建的项目，您可以通过开启 Fluid Compute 来解决此问题，这将最大执行时长延长至 300 秒。
+
+### 开启 Fluid Compute 的步骤（旧版 Vercel 控制台）
+
+1. 前往您在 Vercel 上的项目控制台
+2. 进入 **Settings**（设置）选项卡
+3. 找到 **Functions**（函数）部分
+4. 按照下方截图所示开启 **Fluid Compute**：
+
+![开启 Fluid Compute](https://hub-apac-1.lobeobjects.space/docs/2e2ff332c4b440b584efe1d7ba46aed5.png)
+
+5. 开启后，最大执行时长将默认延长至 300 秒
+
+### 重要说明
+
+- **新项目**：较新的 Vercel 项目默认已启用 Fluid Compute，因此此问题主要影响旧版项目
+
+## 其他资源
+
+有关 Vercel 函数限制和 Fluid Compute 的更多信息：
+
+- [Vercel Fluid Compute 文档](https://vercel.com/docs/fluid-compute)
+- [Vercel 函数限制说明](https://vercel.com/docs/functions/limitations#max-duration)
diff --git a/docs/usage/providers/fal.mdx b/docs/usage/providers/fal.mdx
index 0f3489e1e1..e27ca4ba2c 100644
--- a/docs/usage/providers/fal.mdx
+++ b/docs/usage/providers/fal.mdx
@@ -13,7 +13,7 @@ tags:
 
 # Using Fal in LobeChat
 
-<Image alt={'Using Fal in LobeChat'} cover src={'https://github.com/user-attachments/assets/febb2ffb-8fe8-4f88-b8c9-8fa0985a2352'} />
+<Image alt={'Using Fal in LobeChat'} cover src={'https://hub-apac-1.lobeobjects.space/docs/f253e749baaa2ccac498014178f93091.png'} />
 
 [Fal.ai](https://fal.ai/) is a lightning-fast inference platform specialized in AI media generation, hosting state-of-the-art models for image and video creation including FLUX, Kling, HiDream, and other cutting-edge generative models. This document will guide you on how to use Fal in LobeChat:
 
@@ -28,7 +28,7 @@ tags:
     alt={'Open the creation window'}
     inStep
     src={
-'https://github.com/user-attachments/assets/a2203b3a-1657-485a-a060-b018e7b2faaa'
+'https://hub-apac-1.lobeobjects.space/docs/3f3676e7f9c04a55603bc1174b636b45.png'
 }
   />
 
@@ -36,7 +36,7 @@ tags:
     alt={'Create API Key'}
     inStep
     src={
-'https://github.com/user-attachments/assets/a216e326-6a51-4f3a-b8c1-23bb995ddac2'
+'https://hub-apac-1.lobeobjects.space/docs/214cc5019d9c0810951b33215349136e.png'
 }
   />
 
@@ -44,7 +44,7 @@ tags:
     alt={'Retrieve API Key'}
     inStep
     src={
-'https://github.com/user-attachments/assets/faee998d-4349-4c17-a5c4-07a7ff65f18e'
+'https://hub-apac-1.lobeobjects.space/docs/499a447e98dcc79407d56495d0305e2a.png'
 }
   />
 
@@ -53,12 +53,12 @@ tags:
   - Visit the `Settings` page in LobeChat.
   - Under **AI Service Provider**, locate the **Fal** configuration section.
 
-  <Image alt={'Enter API Key'} inStep src={'https://github.com/user-attachments/assets/7566f679-f935-4a5b-9c98-a8d99d9c7994'} />
+  <Image alt={'Enter API Key'} inStep src={'https://hub-apac-1.lobeobjects.space/docs/fa056feecba0133c76abe1ad12706c05.png'} />
 
   - Paste the API key you obtained.
   - Choose a Fal model (e.g. `fal-ai/flux-pro`, `fal-ai/kling-video`, `fal-ai/hidream-i1-fast`) for image or video generation.
 
-  <Image alt={'Select Fal model for media generation'} inStep src={'https://github.com/user-attachments/assets/817167ff-4723-4f84-8528-c779c5c3a118'} />
+  <Image alt={'Select Fal model for media generation'} inStep src={'https://hub-apac-1.lobeobjects.space/docs/7560502f31b8500032922103fc22e69b.png'} />
 
   <Callout type={'warning'}>
     During usage, you may incur charges according to Fal's pricing policy. Please review Fal's
diff --git a/docs/usage/providers/fal.zh-CN.mdx b/docs/usage/providers/fal.zh-CN.mdx
index 0210c50e69..044f895a94 100644
--- a/docs/usage/providers/fal.zh-CN.mdx
+++ b/docs/usage/providers/fal.zh-CN.mdx
@@ -13,7 +13,7 @@ tags:
 
 # 在 LobeChat 中使用 Fal
 
-<Image alt={'在 LobeChat 中使用 Fal'} cover src={'https://github.com/user-attachments/assets/febb2ffb-8fe8-4f88-b8c9-8fa0985a2352'} />
+<Image alt={'在 LobeChat 中使用 Fal'} cover src={'https://hub-apac-1.lobeobjects.space/docs/f253e749baaa2ccac498014178f93091.png'} />
 
 [Fal.ai](https://fal.ai/) 是一个专门从事 AI 媒体生成的快速推理平台，提供包括 FLUX、Kling、HiDream 等在内的最先进图像和视频生成模型。本文将指导你如何在 LobeChat 中使用 Fal：
 
@@ -28,7 +28,7 @@ tags:
     alt={'打开创建窗口'}
     inStep
     src={
-'https://github.com/user-attachments/assets/a2203b3a-1657-485a-a060-b018e7b2faaa'
+'https://hub-apac-1.lobeobjects.space/docs/3f3676e7f9c04a55603bc1174b636b45.png'
 }
   />
 
@@ -36,7 +36,7 @@ tags:
     alt={'创建 API Key'}
     inStep
     src={
-'https://github.com/user-attachments/assets/a216e326-6a51-4f3a-b8c1-23bb995ddac2'
+'https://hub-apac-1.lobeobjects.space/docs/214cc5019d9c0810951b33215349136e.png'
 }
   />
 
@@ -44,7 +44,7 @@ tags:
     alt={'获取 API Key'}
     inStep
     src={
-'https://github.com/user-attachments/assets/faee998d-4349-4c17-a5c4-07a7ff65f18e'
+'https://hub-apac-1.lobeobjects.space/docs/499a447e98dcc79407d56495d0305e2a.png'
 }
   />
 
@@ -53,12 +53,12 @@ tags:
   - 访问 LobeChat 的 `设置` 页面；
   - 在 `AI服务商` 下找到 `Fal` 的设置项；
 
-  <Image alt={'填入 API 密钥'} inStep src={'https://github.com/user-attachments/assets/7566f679-f935-4a5b-9c98-a8d99d9c7994'} />
+  <Image alt={'填入 API 密钥'} inStep src={'https://hub-apac-1.lobeobjects.space/docs/fa056feecba0133c76abe1ad12706c05.png'} />
 
   - 粘贴获取到的 API Key；
   - 选择一个 Fal 模型（如 `fal-ai/flux-pro`、`fal-ai/kling-video`、`fal-ai/hidream-i1-fast`）用于图像或视频生成。
 
-  <Image alt={'选择 Fal 模型进行媒体生成'} inStep src={'https://github.com/user-attachments/assets/817167ff-4723-4f84-8528-c779c5c3a118'} />
+  <Image alt={'选择 Fal 模型进行媒体生成'} inStep src={'https://hub-apac-1.lobeobjects.space/docs/7560502f31b8500032922103fc22e69b.png'} />
 
   <Callout type={'warning'}>
     在使用过程中，你可能需要向 Fal 支付相应费用，请在大量调用前查阅 Fal 的官方计费政策。
diff --git a/src/libs/model-runtime/utils/openaiCompatibleFactory/index.test.ts b/src/libs/model-runtime/utils/openaiCompatibleFactory/index.test.ts
index 9f372271eb..ded44db6d6 100644
--- a/src/libs/model-runtime/utils/openaiCompatibleFactory/index.test.ts
+++ b/src/libs/model-runtime/utils/openaiCompatibleFactory/index.test.ts
@@ -1392,7 +1392,7 @@ describe('LobeOpenAICompatibleFactory', () => {
         {
           id: 'gemini',
           releasedAt: '2025-01-10',
-          type: undefined,
+          type: 'chat',
         },
       ]);
     });
diff --git a/src/libs/model-runtime/utils/openaiCompatibleFactory/index.ts b/src/libs/model-runtime/utils/openaiCompatibleFactory/index.ts
index 842cdaedc4..01c0e8f3ee 100644
--- a/src/libs/model-runtime/utils/openaiCompatibleFactory/index.ts
+++ b/src/libs/model-runtime/utils/openaiCompatibleFactory/index.ts
@@ -7,6 +7,7 @@ import { Stream } from 'openai/streaming';
 import { LOBE_DEFAULT_MODEL_LIST } from '@/config/aiModels';
 import { RuntimeImageGenParamsValue } from '@/libs/standard-parameters/meta-schema';
 import type { ChatModelCard } from '@/types/llm';
+import { getModelPropertyWithFallback } from '@/utils/getFallbackModelProperty';
 
 import { LobeRuntimeAI } from '../../BaseAI';
 import { AgentRuntimeErrorType, ILobeAgentRuntimeErrorType } from '../../error';
@@ -462,7 +463,7 @@ export const createOpenAICompatibleRuntime = <T extends Record<string, any> = an
       return resultModels.map((model) => {
         return {
           ...model,
-          type: model.type || LOBE_DEFAULT_MODEL_LIST.find((m) => m.id === model.id)?.type,
+          type: model.type || getModelPropertyWithFallback(model.id, 'type'),
         };
       }) as ChatModelCard[];
     }
diff --git a/src/server/globalConfig/genServerAiProviderConfig.test.ts b/src/server/globalConfig/genServerAiProviderConfig.test.ts
new file mode 100644
index 0000000000..879a0411f1
--- /dev/null
+++ b/src/server/globalConfig/genServerAiProviderConfig.test.ts
@@ -0,0 +1,235 @@
+import { beforeEach, describe, expect, it, vi } from 'vitest';
+
+import { ModelProvider } from '@/libs/model-runtime';
+import { AiFullModelCard } from '@/types/aiModel';
+
+import { genServerAiProvidersConfig } from './genServerAiProviderConfig';
+
+// Mock dependencies using importOriginal to preserve real provider data
+vi.mock('@/config/aiModels', async (importOriginal) => {
+  const actual = await importOriginal<typeof import('@/config/aiModels')>();
+  return {
+    ...actual,
+    // Keep the original exports but we can override specific ones if needed
+  };
+});
+
+vi.mock('@/config/llm', () => ({
+  getLLMConfig: vi.fn(() => ({
+    ENABLED_OPENAI: true,
+    ENABLED_ANTHROPIC: false,
+    ENABLED_AI21: false,
+  })),
+}));
+
+vi.mock('@/utils/parseModels', () => ({
+  extractEnabledModels: vi.fn((providerId: string, modelString?: string) => {
+    if (!modelString) return undefined;
+    return [`${providerId}-model-1`, `${providerId}-model-2`];
+  }),
+  transformToAiModelList: vi.fn((params) => {
+    return params.defaultModels;
+  }),
+}));
+
+describe('genServerAiProvidersConfig', () => {
+  beforeEach(() => {
+    vi.clearAllMocks();
+    // Clear environment variables
+    Object.keys(process.env).forEach((key) => {
+      if (key.includes('MODEL_LIST')) {
+        delete process.env[key];
+      }
+    });
+  });
+
+  it('should generate basic provider config with default settings', () => {
+    const result = genServerAiProvidersConfig({});
+
+    expect(result).toHaveProperty('openai');
+    expect(result).toHaveProperty('anthropic');
+
+    expect(result.openai).toEqual({
+      enabled: true,
+      enabledModels: undefined,
+      serverModelLists: expect.any(Array),
+    });
+
+    expect(result.anthropic).toEqual({
+      enabled: false,
+      enabledModels: undefined,
+      serverModelLists: expect.any(Array),
+    });
+  });
+
+  it('should use custom enabled settings from specificConfig', () => {
+    const specificConfig = {
+      openai: {
+        enabled: false,
+      },
+      anthropic: {
+        enabled: true,
+      },
+    };
+
+    const result = genServerAiProvidersConfig(specificConfig);
+
+    expect(result.openai.enabled).toBe(false);
+    expect(result.anthropic.enabled).toBe(true);
+  });
+
+  it('should use custom enabledKey from specificConfig', async () => {
+    const specificConfig = {
+      openai: {
+        enabledKey: 'CUSTOM_OPENAI_ENABLED',
+      },
+    };
+
+    // Mock the LLM config to include our custom key
+    const { getLLMConfig } = vi.mocked(await import('@/config/llm'));
+    getLLMConfig.mockReturnValue({
+      ENABLED_OPENAI: true,
+      ENABLED_ANTHROPIC: false,
+      CUSTOM_OPENAI_ENABLED: true,
+    } as any);
+
+    const result = genServerAiProvidersConfig(specificConfig);
+
+    expect(result.openai.enabled).toBe(true);
+  });
+
+  it('should use environment variables for model lists', async () => {
+    process.env.OPENAI_MODEL_LIST = '+gpt-4,+gpt-3.5-turbo';
+
+    const { extractEnabledModels } = vi.mocked(await import('@/utils/parseModels'));
+    extractEnabledModels.mockReturnValue(['gpt-4', 'gpt-3.5-turbo']);
+
+    const result = genServerAiProvidersConfig({});
+
+    expect(extractEnabledModels).toHaveBeenCalledWith('openai', '+gpt-4,+gpt-3.5-turbo', false);
+    expect(result.openai.enabledModels).toEqual(['gpt-4', 'gpt-3.5-turbo']);
+  });
+
+  it('should use custom modelListKey from specificConfig', async () => {
+    const specificConfig = {
+      openai: {
+        modelListKey: 'CUSTOM_OPENAI_MODELS',
+      },
+    };
+
+    process.env.CUSTOM_OPENAI_MODELS = '+custom-model';
+
+    const { extractEnabledModels } = vi.mocked(await import('@/utils/parseModels'));
+
+    genServerAiProvidersConfig(specificConfig);
+
+    expect(extractEnabledModels).toHaveBeenCalledWith('openai', '+custom-model', false);
+  });
+
+  it('should handle withDeploymentName option', async () => {
+    const specificConfig = {
+      openai: {
+        withDeploymentName: true,
+      },
+    };
+
+    process.env.OPENAI_MODEL_LIST = '+gpt-4->deployment1';
+
+    const { extractEnabledModels, transformToAiModelList } = vi.mocked(
+      await import('@/utils/parseModels'),
+    );
+
+    genServerAiProvidersConfig(specificConfig);
+
+    expect(extractEnabledModels).toHaveBeenCalledWith('openai', '+gpt-4->deployment1', true);
+    expect(transformToAiModelList).toHaveBeenCalledWith({
+      defaultModels: expect.any(Array),
+      modelString: '+gpt-4->deployment1',
+      providerId: 'openai',
+      withDeploymentName: true,
+    });
+  });
+
+  it('should include fetchOnClient when specified in config', () => {
+    const specificConfig = {
+      openai: {
+        fetchOnClient: true,
+      },
+    };
+
+    const result = genServerAiProvidersConfig(specificConfig);
+
+    expect(result.openai).toHaveProperty('fetchOnClient', true);
+  });
+
+  it('should not include fetchOnClient when not specified in config', () => {
+    const result = genServerAiProvidersConfig({});
+
+    expect(result.openai).not.toHaveProperty('fetchOnClient');
+  });
+
+  it('should handle all available providers', () => {
+    const result = genServerAiProvidersConfig({});
+
+    // Check that result includes some key providers
+    expect(result).toHaveProperty('openai');
+    expect(result).toHaveProperty('anthropic');
+
+    // Check structure for each provider
+    Object.keys(result).forEach((provider) => {
+      expect(result[provider]).toHaveProperty('enabled');
+      expect(result[provider]).toHaveProperty('serverModelLists');
+      // enabled can be boolean or undefined (when no config is provided)
+      expect(['boolean', 'undefined']).toContain(typeof result[provider].enabled);
+      expect(Array.isArray(result[provider].serverModelLists)).toBe(true);
+    });
+  });
+});
+
+describe('genServerAiProvidersConfig Error Handling', () => {
+  it('should throw error when a provider is not found in aiModels', async () => {
+    // Reset all mocks to create a clean test environment
+    vi.resetModules();
+
+    // Mock dependencies with a missing provider scenario
+    vi.doMock('@/config/aiModels', () => ({
+      // Explicitly set openai to undefined to simulate missing provider
+      openai: undefined,
+      anthropic: [
+        {
+          id: 'claude-3',
+          displayName: 'Claude 3',
+          type: 'chat',
+          enabled: true,
+        },
+      ],
+    }));
+
+    vi.doMock('@/config/llm', () => ({
+      getLLMConfig: vi.fn(() => ({})),
+    }));
+
+    vi.doMock('@/utils/parseModels', () => ({
+      extractEnabledModels: vi.fn(() => undefined),
+      transformToAiModelList: vi.fn(() => []),
+    }));
+
+    // Mock ModelProvider to include the missing provider
+    vi.doMock('@/libs/model-runtime', () => ({
+      ModelProvider: {
+        openai: 'openai', // This exists in enum
+        anthropic: 'anthropic', // This exists in both enum and aiModels
+      },
+    }));
+
+    // Import the function with the new mocks
+    const { genServerAiProvidersConfig } = await import(
+      './genServerAiProviderConfig?v=' + Date.now()
+    );
+
+    // This should throw because 'openai' is in ModelProvider but not in aiModels
+    expect(() => {
+      genServerAiProvidersConfig({});
+    }).toThrow();
+  });
+});
diff --git a/src/server/globalConfig/genServerAiProviderConfig.ts b/src/server/globalConfig/genServerAiProviderConfig.ts
index 1ae6f6744b..43ad5b8973 100644
--- a/src/server/globalConfig/genServerAiProviderConfig.ts
+++ b/src/server/globalConfig/genServerAiProviderConfig.ts
@@ -3,7 +3,7 @@ import { getLLMConfig } from '@/config/llm';
 import { ModelProvider } from '@/libs/model-runtime';
 import { AiFullModelCard } from '@/types/aiModel';
 import { ProviderConfig } from '@/types/user/settings';
-import { extractEnabledModels, transformToAiChatModelList } from '@/utils/parseModels';
+import { extractEnabledModels, transformToAiModelList } from '@/utils/parseModels';
 
 interface ProviderSpecificConfig {
   enabled?: boolean;
@@ -19,19 +19,17 @@ export const genServerAiProvidersConfig = (specificConfig: Record<any, ProviderS
   return Object.values(ModelProvider).reduce(
     (config, provider) => {
       const providerUpperCase = provider.toUpperCase();
-      const providerCard = AiModels[provider] as AiFullModelCard[];
+      const aiModels = AiModels[provider] as AiFullModelCard[];
 
-      if (!providerCard)
+      if (!aiModels)
         throw new Error(
           `Provider [${provider}] not found in aiModels, please make sure you have exported the provider in the \`aiModels/index.ts\``,
         );
 
       const providerConfig = specificConfig[provider as keyof typeof specificConfig] || {};
-      const providerModelList =
+      const modelString =
         process.env[providerConfig.modelListKey ?? `${providerUpperCase}_MODEL_LIST`];
 
-      const defaultChatModels = providerCard.filter((c) => c.type === 'chat');
-
       config[provider] = {
         enabled:
           typeof providerConfig.enabled !== 'undefined'
@@ -39,12 +37,13 @@ export const genServerAiProvidersConfig = (specificConfig: Record<any, ProviderS
             : llmConfig[providerConfig.enabledKey || `ENABLED_${providerUpperCase}`],
 
         enabledModels: extractEnabledModels(
-          providerModelList,
+          provider,
+          modelString,
           providerConfig.withDeploymentName || false,
         ),
-        serverModelLists: transformToAiChatModelList({
-          defaultChatModels: defaultChatModels || [],
-          modelString: providerModelList,
+        serverModelLists: transformToAiModelList({
+          defaultModels: aiModels || [],
+          modelString,
           providerId: provider,
           withDeploymentName: providerConfig.withDeploymentName || false,
         }),
diff --git a/src/store/aiInfra/slices/aiProvider/action.ts b/src/store/aiInfra/slices/aiProvider/action.ts
index 785949ed8b..ffd3556cdc 100644
--- a/src/store/aiInfra/slices/aiProvider/action.ts
+++ b/src/store/aiInfra/slices/aiProvider/action.ts
@@ -19,6 +19,7 @@ import {
   UpdateAiProviderConfigParams,
   UpdateAiProviderParams,
 } from '@/types/aiProvider';
+import { getModelPropertyWithFallback } from '@/utils/getFallbackModelProperty';
 
 enum AiProviderSwrKey {
   fetchAiProviderItem = 'FETCH_AI_PROVIDER_ITEM',
@@ -216,7 +217,7 @@ export const createAiProviderSlice: StateCreator<
                 ...(model.type === 'image' && {
                   parameters:
                     (model as AIImageModelCard).parameters ||
-                    LOBE_DEFAULT_MODEL_LIST.find((m) => m.id === model.id)?.parameters,
+                    getModelPropertyWithFallback(model.id, 'parameters'),
                 }),
               }));
 
diff --git a/src/utils/getFallbackModelProperty.test.ts b/src/utils/getFallbackModelProperty.test.ts
new file mode 100644
index 0000000000..bf5d3adfb0
--- /dev/null
+++ b/src/utils/getFallbackModelProperty.test.ts
@@ -0,0 +1,193 @@
+import { vi } from 'vitest';
+
+import { getModelPropertyWithFallback } from './getFallbackModelProperty';
+
+// Mock LOBE_DEFAULT_MODEL_LIST for testing
+vi.mock('@/config/aiModels', () => ({
+  LOBE_DEFAULT_MODEL_LIST: [
+    {
+      id: 'gpt-4',
+      providerId: 'openai',
+      type: 'chat',
+      displayName: 'GPT-4',
+      contextWindowTokens: 8192,
+      enabled: true,
+      abilities: {
+        functionCall: true,
+        vision: true,
+      },
+      parameters: {
+        temperature: 0.7,
+        maxTokens: 4096,
+      },
+    },
+    {
+      id: 'gpt-4',
+      providerId: 'azure',
+      type: 'chat',
+      displayName: 'GPT-4 Azure',
+      contextWindowTokens: 8192,
+      enabled: true,
+      abilities: {
+        functionCall: true,
+      },
+    },
+    {
+      id: 'claude-3',
+      providerId: 'anthropic',
+      type: 'chat',
+      displayName: 'Claude 3',
+      contextWindowTokens: 200000,
+      enabled: false,
+    },
+    {
+      id: 'dall-e-3',
+      providerId: 'openai',
+      type: 'image',
+      displayName: 'DALL-E 3',
+      enabled: true,
+      parameters: {
+        size: '1024x1024',
+        quality: 'standard',
+      },
+    },
+  ],
+}));
+
+describe('getModelPropertyWithFallback', () => {
+  describe('when providerId is specified', () => {
+    it('should return exact match value when model exists with specified provider', () => {
+      const result = getModelPropertyWithFallback('gpt-4', 'displayName', 'openai');
+      expect(result).toBe('GPT-4');
+    });
+
+    it('should return exact match type when model exists with specified provider', () => {
+      const result = getModelPropertyWithFallback('gpt-4', 'type', 'openai');
+      expect(result).toBe('chat');
+    });
+
+    it('should return exact match contextWindowTokens when model exists with specified provider', () => {
+      const result = getModelPropertyWithFallback('gpt-4', 'contextWindowTokens', 'azure');
+      expect(result).toBe(8192);
+    });
+
+    it('should fall back to other provider when exact provider match not found', () => {
+      const result = getModelPropertyWithFallback('gpt-4', 'displayName', 'fake-provider');
+      expect(result).toBe('GPT-4'); // Falls back to openai provider
+    });
+
+    it('should return nested property like abilities', () => {
+      const result = getModelPropertyWithFallback('gpt-4', 'abilities', 'openai');
+      expect(result).toEqual({
+        functionCall: true,
+        vision: true,
+      });
+    });
+
+    it('should return parameters property correctly', () => {
+      const result = getModelPropertyWithFallback('dall-e-3', 'parameters', 'openai');
+      expect(result).toEqual({
+        size: '1024x1024',
+        quality: 'standard',
+      });
+    });
+  });
+
+  describe('when providerId is not specified', () => {
+    it('should return fallback match value when model exists', () => {
+      const result = getModelPropertyWithFallback('claude-3', 'displayName');
+      expect(result).toBe('Claude 3');
+    });
+
+    it('should return fallback match type when model exists', () => {
+      const result = getModelPropertyWithFallback('claude-3', 'type');
+      expect(result).toBe('chat');
+    });
+
+    it('should return fallback match enabled property', () => {
+      const result = getModelPropertyWithFallback('claude-3', 'enabled');
+      expect(result).toBe(false);
+    });
+  });
+
+  describe('when model is not found', () => {
+    it('should return default value "chat" for type property', () => {
+      const result = getModelPropertyWithFallback('non-existent-model', 'type');
+      expect(result).toBe('chat');
+    });
+
+    it('should return default value "chat" for type property even with providerId', () => {
+      const result = getModelPropertyWithFallback('non-existent-model', 'type', 'fake-provider');
+      expect(result).toBe('chat');
+    });
+
+    it('should return undefined for non-type properties when model not found', () => {
+      const result = getModelPropertyWithFallback('non-existent-model', 'displayName');
+      expect(result).toBeUndefined();
+    });
+
+    it('should return undefined for contextWindowTokens when model not found', () => {
+      const result = getModelPropertyWithFallback('non-existent-model', 'contextWindowTokens');
+      expect(result).toBeUndefined();
+    });
+
+    it('should return undefined for enabled property when model not found', () => {
+      const result = getModelPropertyWithFallback('non-existent-model', 'enabled');
+      expect(result).toBeUndefined();
+    });
+  });
+
+  describe('provider precedence logic', () => {
+    it('should prioritize exact provider match over general match', () => {
+      // gpt-4 exists in both openai and azure providers with different displayNames
+      const openaiResult = getModelPropertyWithFallback('gpt-4', 'displayName', 'openai');
+      const azureResult = getModelPropertyWithFallback('gpt-4', 'displayName', 'azure');
+
+      expect(openaiResult).toBe('GPT-4');
+      expect(azureResult).toBe('GPT-4 Azure');
+    });
+
+    it('should fall back to first match when specified provider not found', () => {
+      // When asking for 'fake-provider', should fall back to first match (openai)
+      const result = getModelPropertyWithFallback('gpt-4', 'displayName', 'fake-provider');
+      expect(result).toBe('GPT-4');
+    });
+  });
+
+  describe('property existence handling', () => {
+    it('should handle undefined properties gracefully', () => {
+      // claude-3 doesn't have abilities property defined
+      const result = getModelPropertyWithFallback('claude-3', 'abilities');
+      expect(result).toBeUndefined();
+    });
+
+    it('should handle properties that exist but have falsy values', () => {
+      // claude-3 has enabled: false
+      const result = getModelPropertyWithFallback('claude-3', 'enabled');
+      expect(result).toBe(false);
+    });
+
+    it('should distinguish between undefined and null values', () => {
+      // Testing that we check for undefined specifically, not just falsy values
+      const result = getModelPropertyWithFallback('claude-3', 'contextWindowTokens');
+      expect(result).toBe(200000); // Should find the defined value
+    });
+  });
+
+  describe('edge cases', () => {
+    it('should handle empty string modelId', () => {
+      const result = getModelPropertyWithFallback('', 'type');
+      expect(result).toBe('chat'); // Should fall back to default
+    });
+
+    it('should handle empty string providerId', () => {
+      const result = getModelPropertyWithFallback('gpt-4', 'type', '');
+      expect(result).toBe('chat'); // Should still find the model via fallback
+    });
+
+    it('should handle case-sensitive modelId correctly', () => {
+      const result = getModelPropertyWithFallback('GPT-4', 'type'); // Wrong case
+      expect(result).toBe('chat'); // Should fall back to default since no match
+    });
+  });
+});
diff --git a/src/utils/getFallbackModelProperty.ts b/src/utils/getFallbackModelProperty.ts
new file mode 100644
index 0000000000..9de3621c62
--- /dev/null
+++ b/src/utils/getFallbackModelProperty.ts
@@ -0,0 +1,36 @@
+import { LOBE_DEFAULT_MODEL_LIST } from '@/config/aiModels';
+import { AiFullModelCard } from '@/types/aiModel';
+
+/**
+ * Get the model property value, first from the specified provider, and then from other providers as a fallback.
+ * @param modelId The ID of the model.
+ * @param propertyName The name of the property.
+ * @param providerId Optional provider ID for an exact match.
+ * @returns The property value or a default value.
+ */
+export const getModelPropertyWithFallback = <T>(
+  modelId: string,
+  propertyName: keyof AiFullModelCard,
+  providerId?: string,
+): T => {
+  // Step 1: If providerId is provided, prioritize an exact match (same provider + same id)
+  if (providerId) {
+    const exactMatch = LOBE_DEFAULT_MODEL_LIST.find(
+      (m) => m.id === modelId && m.providerId === providerId,
+    );
+
+    if (exactMatch && exactMatch[propertyName] !== undefined) {
+      return exactMatch[propertyName] as T;
+    }
+  }
+
+  // Step 2: Fallback to a match ignoring the provider (match id only)
+  const fallbackMatch = LOBE_DEFAULT_MODEL_LIST.find((m) => m.id === modelId);
+
+  if (fallbackMatch && fallbackMatch[propertyName] !== undefined) {
+    return fallbackMatch[propertyName] as T;
+  }
+
+  // Step 3: Return a default value
+  return (propertyName === 'type' ? 'chat' : undefined) as T;
+};
diff --git a/src/utils/parseModels.test.ts b/src/utils/parseModels.test.ts
index 3e42a911b1..806bdd93eb 100644
--- a/src/utils/parseModels.test.ts
+++ b/src/utils/parseModels.test.ts
@@ -4,11 +4,12 @@ import { LOBE_DEFAULT_MODEL_LIST } from '@/config/aiModels';
 import { openaiChatModels } from '@/config/aiModels/openai';
 import { AiFullModelCard } from '@/types/aiModel';
 
-import { parseModelString, transformToAiChatModelList } from './parseModels';
+import { extractEnabledModels, parseModelString, transformToAiModelList } from './parseModels';
 
 describe('parseModelString', () => {
   it('custom deletion, addition, and renaming of models', () => {
     const result = parseModelString(
+      'test-provider',
       '-all,+llama,+claude-2，-gpt-3.5-turbo,gpt-4-1106-preview=gpt-4-turbo,gpt-4-1106-preview=gpt-4-32k',
     );
 
@@ -16,24 +17,30 @@ describe('parseModelString', () => {
   });
 
   it('duplicate naming model', () => {
-    const result = parseModelString('gpt-4-1106-preview=gpt-4-turbo，gpt-4-1106-preview=gpt-4-32k');
+    const result = parseModelString(
+      'test-provider',
+      'gpt-4-1106-preview=gpt-4-turbo，gpt-4-1106-preview=gpt-4-32k',
+    );
     expect(result).toMatchSnapshot();
   });
 
   it('only add the model', () => {
-    const result = parseModelString('model1,model2,model3，model4');
+    const result = parseModelString('test-provider', 'model1,model2,model3，model4');
 
     expect(result).toMatchSnapshot();
   });
 
   it('empty string model', () => {
-    const result = parseModelString('gpt-4-1106-preview=gpt-4-turbo,,  ,\n  ，+claude-2');
+    const result = parseModelString(
+      'test-provider',
+      'gpt-4-1106-preview=gpt-4-turbo,,  ,\n  ，+claude-2',
+    );
     expect(result).toMatchSnapshot();
   });
 
   describe('extension capabilities', () => {
     it('with token', () => {
-      const result = parseModelString('chatglm-6b=ChatGLM 6B<4096>');
+      const result = parseModelString('test-provider', 'chatglm-6b=ChatGLM 6B<4096>');
 
       expect(result.add[0]).toEqual({
         displayName: 'ChatGLM 6B',
@@ -45,7 +52,7 @@ describe('parseModelString', () => {
     });
 
     it('token and function calling', () => {
-      const result = parseModelString('spark-v3.5=讯飞星火 v3.5<8192:fc>');
+      const result = parseModelString('test-provider', 'spark-v3.5=讯飞星火 v3.5<8192:fc>');
 
       expect(result.add[0]).toEqual({
         displayName: '讯飞星火 v3.5',
@@ -59,7 +66,7 @@ describe('parseModelString', () => {
     });
 
     it('token and reasoning', () => {
-      const result = parseModelString('deepseek-r1=Deepseek R1<65536:reasoning>');
+      const result = parseModelString('test-provider', 'deepseek-r1=Deepseek R1<65536:reasoning>');
 
       expect(result.add[0]).toEqual({
         displayName: 'Deepseek R1',
@@ -73,7 +80,7 @@ describe('parseModelString', () => {
     });
 
     it('token and search', () => {
-      const result = parseModelString('qwen-max-latest=Qwen Max<32768:search>');
+      const result = parseModelString('test-provider', 'qwen-max-latest=Qwen Max<32768:search>');
 
       expect(result.add[0]).toEqual({
         displayName: 'Qwen Max',
@@ -88,6 +95,7 @@ describe('parseModelString', () => {
 
     it('token and image output', () => {
       const result = parseModelString(
+        'test-provider',
         'gemini-2.0-flash-exp-image-generation=Gemini 2.0 Flash (Image Generation) Experimental<32768:imageOutput>',
       );
 
@@ -104,6 +112,7 @@ describe('parseModelString', () => {
 
     it('multi models', () => {
       const result = parseModelString(
+        'test-provider',
         'gemini-1.5-flash-latest=Gemini 1.5 Flash<16000:vision>,gpt-4-all=ChatGPT Plus<128000:fc:vision:file>',
       );
 
@@ -133,6 +142,7 @@ describe('parseModelString', () => {
 
     it('should have file with builtin models like gpt-4-0125-preview', () => {
       const result = parseModelString(
+        'openai',
         '-all,+gpt-4-0125-preview=ChatGPT-4<128000:fc:file>,+gpt-4-turbo-2024-04-09=ChatGPT-4 Vision<128000:fc:vision:file>',
       );
       expect(result.add).toEqual([
@@ -161,7 +171,7 @@ describe('parseModelString', () => {
     });
 
     it('should handle empty extension capability value', () => {
-      const result = parseModelString('model1<1024:>');
+      const result = parseModelString('test-provider', 'model1<1024:>');
       expect(result.add[0]).toEqual({
         abilities: {},
         type: 'chat',
@@ -171,7 +181,7 @@ describe('parseModelString', () => {
     });
 
     it('should handle empty extension capability name', () => {
-      const result = parseModelString('model1<1024::file>');
+      const result = parseModelString('test-provider', 'model1<1024::file>');
       expect(result.add[0]).toEqual({
         id: 'model1',
         contextWindowTokens: 1024,
@@ -183,7 +193,7 @@ describe('parseModelString', () => {
     });
 
     it('should handle duplicate extension capabilities', () => {
-      const result = parseModelString('model1<1024:vision:vision>');
+      const result = parseModelString('test-provider', 'model1<1024:vision:vision>');
       expect(result.add[0]).toEqual({
         id: 'model1',
         contextWindowTokens: 1024,
@@ -195,7 +205,7 @@ describe('parseModelString', () => {
     });
 
     it('should handle case-sensitive extension capability names', () => {
-      const result = parseModelString('model1<1024:VISION:FC:file>');
+      const result = parseModelString('test-provider', 'model1<1024:VISION:FC:file>');
       expect(result.add[0]).toEqual({
         id: 'model1',
         contextWindowTokens: 1024,
@@ -207,7 +217,7 @@ describe('parseModelString', () => {
     });
 
     it('should handle case-sensitive extension capability values', () => {
-      const result = parseModelString('model1<1024:vision:Fc:File>');
+      const result = parseModelString('test-provider', 'model1<1024:vision:Fc:File>');
       expect(result.add[0]).toEqual({
         id: 'model1',
         contextWindowTokens: 1024,
@@ -219,12 +229,12 @@ describe('parseModelString', () => {
     });
 
     it('should handle empty angle brackets', () => {
-      const result = parseModelString('model1<>');
+      const result = parseModelString('test-provider', 'model1<>');
       expect(result.add[0]).toEqual({ id: 'model1', abilities: {}, type: 'chat' });
     });
 
     it('should handle not close angle brackets', () => {
-      const result = parseModelString('model1<,model2');
+      const result = parseModelString('test-provider', 'model1<,model2');
       expect(result.add).toEqual([
         { id: 'model1', abilities: {}, type: 'chat' },
         { id: 'model2', abilities: {}, type: 'chat' },
@@ -232,7 +242,7 @@ describe('parseModelString', () => {
     });
 
     it('should handle multi close angle brackets', () => {
-      const result = parseModelString('model1<>>,model2');
+      const result = parseModelString('test-provider', 'model1<>>,model2');
       expect(result.add).toEqual([
         { id: 'model1', abilities: {}, type: 'chat' },
         { id: 'model2', abilities: {}, type: 'chat' },
@@ -240,22 +250,22 @@ describe('parseModelString', () => {
     });
 
     it('should handle only colon inside angle brackets', () => {
-      const result = parseModelString('model1<:>');
+      const result = parseModelString('test-provider', 'model1<:>');
       expect(result.add[0]).toEqual({ id: 'model1', abilities: {}, type: 'chat' });
     });
 
     it('should handle only non-digit characters inside angle brackets', () => {
-      const result = parseModelString('model1<abc>');
+      const result = parseModelString('test-provider', 'model1<abc>');
       expect(result.add[0]).toEqual({ id: 'model1', abilities: {}, type: 'chat' });
     });
 
     it('should handle non-digit characters followed by digits inside angle brackets', () => {
-      const result = parseModelString('model1<abc123>');
+      const result = parseModelString('test-provider', 'model1<abc123>');
       expect(result.add[0]).toEqual({ id: 'model1', abilities: {}, type: 'chat' });
     });
 
     it('should handle digits followed by non-colon characters inside angle brackets', () => {
-      const result = parseModelString('model1<1024abc>');
+      const result = parseModelString('test-provider', 'model1<1024abc>');
       expect(result.add[0]).toEqual({
         id: 'model1',
         contextWindowTokens: 1024,
@@ -265,7 +275,7 @@ describe('parseModelString', () => {
     });
 
     it('should handle digits followed by multiple colons inside angle brackets', () => {
-      const result = parseModelString('model1<1024::>');
+      const result = parseModelString('test-provider', 'model1<1024::>');
       expect(result.add[0]).toEqual({
         id: 'model1',
         contextWindowTokens: 1024,
@@ -275,7 +285,7 @@ describe('parseModelString', () => {
     });
 
     it('should handle digits followed by a colon and non-letter characters inside angle brackets', () => {
-      const result = parseModelString('model1<1024:123>');
+      const result = parseModelString('test-provider', 'model1<1024:123>');
       expect(result.add[0]).toEqual({
         id: 'model1',
         contextWindowTokens: 1024,
@@ -285,7 +295,7 @@ describe('parseModelString', () => {
     });
 
     it('should handle digits followed by a colon and spaces inside angle brackets', () => {
-      const result = parseModelString('model1<1024: vision>');
+      const result = parseModelString('test-provider', 'model1<1024: vision>');
       expect(result.add[0]).toEqual({
         id: 'model1',
         contextWindowTokens: 1024,
@@ -295,7 +305,7 @@ describe('parseModelString', () => {
     });
 
     it('should handle digits followed by multiple colons and spaces inside angle brackets', () => {
-      const result = parseModelString('model1<1024: : vision>');
+      const result = parseModelString('test-provider', 'model1<1024: : vision>');
       expect(result.add[0]).toEqual({
         id: 'model1',
         contextWindowTokens: 1024,
@@ -305,9 +315,64 @@ describe('parseModelString', () => {
     });
   });
 
+  describe('FAL image models', () => {
+    it('should correctly parse FAL image model ids with slash and custom display names', () => {
+      const result = parseModelString(
+        'fal',
+        '-all,+flux-kontext/dev=KontextDev,+flux-pro/kontext=KontextPro,+flux/schnell=Schnell,+imagen4/preview=Imagen4',
+      );
+      expect(result.add).toEqual([
+        {
+          id: 'flux-kontext/dev',
+          displayName: 'KontextDev',
+          abilities: {},
+          type: 'image',
+        },
+        {
+          id: 'flux-pro/kontext',
+          displayName: 'KontextPro',
+          abilities: {},
+          type: 'image',
+        },
+        {
+          id: 'flux/schnell',
+          displayName: 'Schnell',
+          abilities: {},
+          type: 'image',
+        },
+        {
+          id: 'imagen4/preview',
+          displayName: 'Imagen4',
+          abilities: {},
+          type: 'image',
+        },
+      ]);
+      expect(result.removeAll).toBe(true);
+      expect(result.removed).toEqual(['all']);
+    });
+
+    it('should correctly parse FAL image model ids with slash (no displayName)', () => {
+      const result = parseModelString('fal', '-all,+flux-kontext/dev,+flux-pro/kontext');
+      expect(result.add).toEqual([
+        {
+          id: 'flux-kontext/dev',
+          abilities: {},
+          type: 'image',
+        },
+        {
+          id: 'flux-pro/kontext',
+          abilities: {},
+          type: 'image',
+        },
+      ]);
+      expect(result.removeAll).toBe(true);
+      expect(result.removed).toEqual(['all']);
+    });
+  });
+
   describe('deployment name', () => {
     it('should have no deployment name', () => {
-      const result = parseModelString('model1=Model 1', true);
+      const result = parseModelString('test-provider', 'model1=Model 1', true);
       expect(result.add[0]).toEqual({
         id: 'model1',
         displayName: 'Model 1',
@@ -317,7 +382,7 @@ describe('parseModelString', () => {
     });
 
     it('should have diff deployment name as id', () => {
-      const result = parseModelString('gpt-35-turbo->my-deploy=GPT 3.5 Turbo', true);
+      const result = parseModelString('azure', 'gpt-35-turbo->my-deploy=GPT 3.5 Turbo', true);
       expect(result.add[0]).toEqual({
         id: 'gpt-35-turbo',
         displayName: 'GPT 3.5 Turbo',
@@ -331,6 +396,7 @@ describe('parseModelString', () => {
 
     it('should handle with multi deployName', () => {
       const result = parseModelString(
+        'azure',
         'gpt-4o->id1=GPT-4o,gpt-4o-mini->id2=gpt-4o-mini,o1-mini->id3=O1 mini',
         true,
       );
@@ -361,6 +427,42 @@ describe('parseModelString', () => {
   });
 });
 
+describe('extractEnabledModels', () => {
+  it('should return undefined when no models are added', () => {
+    const result = extractEnabledModels('test-provider', '-all');
+    expect(result).toBeUndefined();
+  });
+
+  it('should return undefined when modelString is empty', () => {
+    const result = extractEnabledModels('test-provider', '');
+    expect(result).toBeUndefined();
+  });
+
+  it('should return array of model IDs when models are added', () => {
+    const result = extractEnabledModels('test-provider', '+model1,+model2,+model3');
+    expect(result).toEqual(['model1', 'model2', 'model3']);
+  });
+
+  it('should handle mixed add/remove operations and return only added models', () => {
+    const result = extractEnabledModels('test-provider', '+model1,-model2,+model3');
+    expect(result).toEqual(['model1', 'model3']);
+  });
+
+  it('should handle deployment names when withDeploymentName is true', () => {
+    const result = extractEnabledModels(
+      'azure',
+      '+gpt-4->deployment1,+gpt-35-turbo->deployment2',
+      true,
+    );
+    expect(result).toEqual(['gpt-4', 'gpt-35-turbo']);
+  });
+
+  it('should handle complex model strings with custom names', () => {
+    const result = extractEnabledModels('openai', '+gpt-4=Custom GPT-4,+claude-2=Custom Claude');
+    expect(result).toEqual(['gpt-4', 'claude-2']);
+  });
+});
+
 describe('transformToChatModelCards', () => {
   const defaultChatModels: AiFullModelCard[] = [
     { id: 'model1', displayName: 'Model 1', enabled: true, type: 'chat' },
@@ -368,27 +470,27 @@ describe('transformToChatModelCards', () => {
   ];
 
   it('should return undefined when modelString is empty', () => {
-    const result = transformToAiChatModelList({
+    const result = transformToAiModelList({
       modelString: '',
-      defaultChatModels,
+      defaultModels: defaultChatModels,
       providerId: 'openai',
     });
     expect(result).toBeUndefined();
   });
 
   it('should remove all models when removeAll is true', () => {
-    const result = transformToAiChatModelList({
+    const result = transformToAiModelList({
       modelString: '-all',
-      defaultChatModels,
+      defaultModels: defaultChatModels,
       providerId: 'openai',
     });
     expect(result).toEqual([]);
   });
 
   it('should remove specified models', () => {
-    const result = transformToAiChatModelList({
+    const result = transformToAiModelList({
       modelString: '-model1',
-      defaultChatModels,
+      defaultModels: defaultChatModels,
       providerId: 'openai',
     });
     expect(result).toEqual([
@@ -398,9 +500,9 @@ describe('transformToChatModelCards', () => {
 
   it('should add a new known model', () => {
     const knownModel = LOBE_DEFAULT_MODEL_LIST.find((m) => m.providerId === 'ai21')!;
-    const result = transformToAiChatModelList({
+    const result = transformToAiModelList({
       modelString: `${knownModel.id}`,
-      defaultChatModels,
+      defaultModels: defaultChatModels,
       providerId: 'ai21',
     });
 
@@ -413,9 +515,9 @@ describe('transformToChatModelCards', () => {
 
   it('should update an existing known model', () => {
     const knownModel = LOBE_DEFAULT_MODEL_LIST.find((m) => m.providerId === 'openai')!;
-    const result = transformToAiChatModelList({
+    const result = transformToAiModelList({
       modelString: `+${knownModel.id}=Updated Model`,
-      defaultChatModels: [knownModel],
+      defaultModels: [knownModel],
       providerId: 'openai',
     });
 
@@ -427,9 +529,9 @@ describe('transformToChatModelCards', () => {
   });
 
   it('should add a new custom model', () => {
-    const result = transformToAiChatModelList({
+    const result = transformToAiModelList({
       modelString: '+custom_model=Custom Model',
-      defaultChatModels,
+      defaultModels: defaultChatModels,
       providerId: 'openai',
     });
     expect(result).toContainEqual({
@@ -442,10 +544,10 @@ describe('transformToChatModelCards', () => {
   });
 
   it('should have file with builtin models like gpt-4-0125-preview', () => {
-    const result = transformToAiChatModelList({
+    const result = transformToAiModelList({
       modelString:
         '-all,+gpt-4-0125-preview=ChatGPT-4<128000:fc:file>,+gpt-4-turbo-2024-04-09=ChatGPT-4 Vision<128000:fc:vision:file>',
-      defaultChatModels: openaiChatModels,
+      defaultModels: openaiChatModels,
       providerId: 'openai',
     });
 
@@ -457,9 +559,9 @@ describe('transformToChatModelCards', () => {
       (m) => m.id === 'deepseek-r1' && m.providerId === 'volcengine',
     );
     const defaultChatModels: AiFullModelCard[] = [];
-    const result = transformToAiChatModelList({
+    const result = transformToAiModelList({
       modelString: '+deepseek-r1',
-      defaultChatModels,
+      defaultModels: defaultChatModels,
       providerId: 'volcengine',
       withDeploymentName: true,
     });
@@ -474,9 +576,9 @@ describe('transformToChatModelCards', () => {
     const knownModel = LOBE_DEFAULT_MODEL_LIST.find(
       (m) => m.id === 'deepseek-r1' && m.providerId === 'volcengine',
     );
-    const result = transformToAiChatModelList({
+    const result = transformToAiModelList({
       modelString: `+deepseek-r1->my-custom-deploy`,
-      defaultChatModels,
+      defaultModels: defaultChatModels,
       providerId: 'volcengine',
       withDeploymentName: true,
     });
@@ -489,9 +591,9 @@ describe('transformToChatModelCards', () => {
 
   it('should set both id and deploymentName to the full string when no -> is used and withDeploymentName is true', () => {
     const defaultChatModels: AiFullModelCard[] = [];
-    const result = transformToAiChatModelList({
+    const result = transformToAiModelList({
       modelString: `+my_model`,
-      defaultChatModels,
+      defaultModels: defaultChatModels,
       providerId: 'volcengine',
       withDeploymentName: true,
     });
@@ -602,9 +704,9 @@ describe('transformToChatModelCards', () => {
     const modelString =
       '-all,gpt-4o->id1=GPT-4o,gpt-4o-mini->id2=GPT 4o Mini,o1-mini->id3=OpenAI o1-mini';
 
-    const data = transformToAiChatModelList({
+    const data = transformToAiModelList({
       modelString,
-      defaultChatModels,
+      defaultModels: defaultChatModels,
       providerId: 'azure',
       withDeploymentName: true,
     });
diff --git a/src/utils/parseModels.ts b/src/utils/parseModels.ts
index 1e9fbad34e..d765a91380 100644
--- a/src/utils/parseModels.ts
+++ b/src/utils/parseModels.ts
@@ -1,13 +1,18 @@
 import { produce } from 'immer';
 
 import { LOBE_DEFAULT_MODEL_LIST } from '@/config/aiModels';
-import { AiFullModelCard } from '@/types/aiModel';
+import { AiFullModelCard, AiModelType } from '@/types/aiModel';
+import { getModelPropertyWithFallback } from '@/utils/getFallbackModelProperty';
 import { merge } from '@/utils/merge';
 
 /**
  * Parse model string to add or remove models.
  */
-export const parseModelString = (modelString: string = '', withDeploymentName = false) => {
+export const parseModelString = (
+  providerId: string,
+  modelString: string = '',
+  withDeploymentName = false,
+) => {
   let models: AiFullModelCard[] = [];
   let removeAll = false;
   const removedModels: string[] = [];
@@ -46,12 +51,18 @@ export const parseModelString = (modelString: string = '', withDeploymentName =
       models.splice(existingIndex, 1);
     }
 
+    // Use new type lookup function, prioritizing same provider first, then fallback to other providers
+    const modelType: AiModelType = getModelPropertyWithFallback<AiModelType>(
+      id,
+      'type',
+      providerId,
+    );
+
     const model: AiFullModelCard = {
       abilities: {},
       displayName: displayName || undefined,
       id,
-      // TODO: 临时写死为 chat ，后续基于元数据迭代成对应的类型
-      type: 'chat',
+      type: modelType,
     };
 
     if (deploymentName) {
@@ -108,21 +119,21 @@ export const parseModelString = (modelString: string = '', withDeploymentName =
 /**
  * Extract a special method to process chatModels
  */
-export const transformToAiChatModelList = ({
+export const transformToAiModelList = ({
   modelString = '',
-  defaultChatModels,
+  defaultModels,
   providerId,
   withDeploymentName = false,
 }: {
-  defaultChatModels: AiFullModelCard[];
+  defaultModels: AiFullModelCard[];
   modelString?: string;
   providerId: string;
   withDeploymentName?: boolean;
 }): AiFullModelCard[] | undefined => {
   if (!modelString) return undefined;
 
-  const modelConfig = parseModelString(modelString, withDeploymentName);
-  let chatModels = modelConfig.removeAll ? [] : defaultChatModels;
+  const modelConfig = parseModelString(providerId, modelString, withDeploymentName);
+  let chatModels = modelConfig.removeAll ? [] : defaultModels;
 
   // 处理移除逻辑
   if (!modelConfig.removeAll) {
@@ -182,8 +193,12 @@ export const transformToAiChatModelList = ({
   });
 };
 
-export const extractEnabledModels = (modelString: string = '', withDeploymentName = false) => {
-  const modelConfig = parseModelString(modelString, withDeploymentName);
+export const extractEnabledModels = (
+  providerId: string,
+  modelString: string = '',
+  withDeploymentName = false,
+) => {
+  const modelConfig = parseModelString(providerId, modelString, withDeploymentName);
   const list = modelConfig.add.map((m) => m.id);
 
   if (list.length === 0) return;