diff --git a/packages/core/src/config/models.ts b/packages/core/src/config/models.ts
index 15ca9cf1..27d29053 100644
--- a/packages/core/src/config/models.ts
+++ b/packages/core/src/config/models.ts
@@ -4,6 +4,9 @@
  * SPDX-License-Identifier: Apache-2.0
  */
 
+export const DEFAULT_QWEN_MODEL = 'qwen3-coder-plus';
+export const DEFAULT_QWEN_FLASH_MODEL = 'qwen3-coder-flash';
+
 export const DEFAULT_GEMINI_MODEL = 'qwen3-coder-plus';
 export const DEFAULT_GEMINI_FLASH_MODEL = 'gemini-2.5-flash';
 export const DEFAULT_GEMINI_FLASH_LITE_MODEL = 'gemini-2.5-flash-lite';
diff --git a/packages/core/src/core/contentGenerator.ts b/packages/core/src/core/contentGenerator.ts
index c4a934e7..b04ae36d 100644
--- a/packages/core/src/core/contentGenerator.ts
+++ b/packages/core/src/core/contentGenerator.ts
@@ -14,7 +14,7 @@ import {
   GoogleGenAI,
 } from '@google/genai';
 import { createCodeAssistContentGenerator } from '../code_assist/codeAssist.js';
-import { DEFAULT_GEMINI_MODEL } from '../config/models.js';
+import { DEFAULT_GEMINI_MODEL, DEFAULT_QWEN_MODEL } from '../config/models.js';
 import { Config } from '../config/config.js';
 import { getEffectiveModel } from './modelCheck.js';
 import { UserTierId } from '../code_assist/types.js';
@@ -136,7 +136,9 @@ export function createContentGeneratorConfig(
     // For Qwen OAuth, we'll handle the API key dynamically in createContentGenerator
     // Set a special marker to indicate this is Qwen OAuth
     contentGeneratorConfig.apiKey = 'QWEN_OAUTH_DYNAMIC_TOKEN';
-    contentGeneratorConfig.model = config.getModel() || DEFAULT_GEMINI_MODEL;
+
+    // Prefer to use qwen3-coder-plus as the default Qwen model if QWEN_MODEL is not set.
+    contentGeneratorConfig.model = process.env.QWEN_MODEL || DEFAULT_QWEN_MODEL;
 
     return contentGeneratorConfig;
   }
diff --git a/packages/core/src/core/openaiContentGenerator.test.ts b/packages/core/src/core/openaiContentGenerator.test.ts
index 6dc0690a..84ba2dcb 100644
--- a/packages/core/src/core/openaiContentGenerator.test.ts
+++ b/packages/core/src/core/openaiContentGenerator.test.ts
@@ -2384,4 +2384,605 @@ describe('OpenAIContentGenerator', () => {
       consoleSpy.mockRestore();
     });
   });
+
+  describe('metadata control', () => {
+    it('should include metadata when authType is QWEN_OAUTH', async () => {
+      const qwenConfig = {
+        getContentGeneratorConfig: vi.fn().mockReturnValue({
+          authType: 'qwen-oauth',
+          enableOpenAILogging: false,
+        }),
+        getSessionId: vi.fn().mockReturnValue('test-session-id'),
+      } as unknown as Config;
+
+      const qwenGenerator = new OpenAIContentGenerator(
+        'test-key',
+        'qwen-turbo',
+        qwenConfig,
+      );
+
+      const mockResponse = {
+        id: 'chatcmpl-123',
+        choices: [
+          {
+            index: 0,
+            message: { role: 'assistant', content: 'Response' },
+            finish_reason: 'stop',
+          },
+        ],
+        created: 1677652288,
+        model: 'qwen-turbo',
+      };
+
+      mockOpenAIClient.chat.completions.create.mockResolvedValue(mockResponse);
+
+      const request: GenerateContentParameters = {
+        contents: [{ role: 'user', parts: [{ text: 'Hello' }] }],
+        model: 'qwen-turbo',
+      };
+
+      await qwenGenerator.generateContent(request, 'test-prompt-id');
+
+      expect(mockOpenAIClient.chat.completions.create).toHaveBeenCalledWith(
+        expect.objectContaining({
+          metadata: {
+            sessionId: 'test-session-id',
+            promptId: 'test-prompt-id',
+          },
+        }),
+      );
+    });
+
+    it('should include metadata when baseURL is dashscope openai compatible mode', async () => {
+      // Mock environment to set dashscope base URL BEFORE creating the generator
+      vi.stubEnv(
+        'OPENAI_BASE_URL',
+        'https://dashscope.aliyuncs.com/compatible-mode/v1',
+      );
+
+      const dashscopeConfig = {
+        getContentGeneratorConfig: vi.fn().mockReturnValue({
+          authType: 'openai', // Not QWEN_OAUTH
+          enableOpenAILogging: false,
+        }),
+        getSessionId: vi.fn().mockReturnValue('dashscope-session-id'),
+      } as unknown as Config;
+
+      const dashscopeGenerator = new OpenAIContentGenerator(
+        'test-key',
+        'qwen-turbo',
+        dashscopeConfig,
+      );
+
+      // Debug: Check if the client was created with the correct baseURL
+      expect(vi.mocked(OpenAI)).toHaveBeenCalledWith(
+        expect.objectContaining({
+          baseURL: 'https://dashscope.aliyuncs.com/compatible-mode/v1',
+        }),
+      );
+
+      // Mock the client's baseURL property to return the expected value
+      Object.defineProperty(dashscopeGenerator['client'], 'baseURL', {
+        value: 'https://dashscope.aliyuncs.com/compatible-mode/v1',
+        writable: true,
+      });
+
+      const mockResponse = {
+        id: 'chatcmpl-123',
+        choices: [
+          {
+            index: 0,
+            message: { role: 'assistant', content: 'Response' },
+            finish_reason: 'stop',
+          },
+        ],
+        created: 1677652288,
+        model: 'qwen-turbo',
+      };
+
+      mockOpenAIClient.chat.completions.create.mockResolvedValue(mockResponse);
+
+      const request: GenerateContentParameters = {
+        contents: [{ role: 'user', parts: [{ text: 'Hello' }] }],
+        model: 'qwen-turbo',
+      };
+
+      await dashscopeGenerator.generateContent(request, 'dashscope-prompt-id');
+
+      expect(mockOpenAIClient.chat.completions.create).toHaveBeenCalledWith(
+        expect.objectContaining({
+          metadata: {
+            sessionId: 'dashscope-session-id',
+            promptId: 'dashscope-prompt-id',
+          },
+        }),
+      );
+    });
+
+    it('should NOT include metadata for regular OpenAI providers', async () => {
+      const regularConfig = {
+        getContentGeneratorConfig: vi.fn().mockReturnValue({
+          authType: 'openai',
+          enableOpenAILogging: false,
+        }),
+        getSessionId: vi.fn().mockReturnValue('regular-session-id'),
+      } as unknown as Config;
+
+      const regularGenerator = new OpenAIContentGenerator(
+        'test-key',
+        'gpt-4',
+        regularConfig,
+      );
+
+      const mockResponse = {
+        id: 'chatcmpl-123',
+        choices: [
+          {
+            index: 0,
+            message: { role: 'assistant', content: 'Response' },
+            finish_reason: 'stop',
+          },
+        ],
+        created: 1677652288,
+        model: 'gpt-4',
+      };
+
+      mockOpenAIClient.chat.completions.create.mockResolvedValue(mockResponse);
+
+      const request: GenerateContentParameters = {
+        contents: [{ role: 'user', parts: [{ text: 'Hello' }] }],
+        model: 'gpt-4',
+      };
+
+      await regularGenerator.generateContent(request, 'regular-prompt-id');
+
+      // Should NOT include metadata
+      expect(mockOpenAIClient.chat.completions.create).toHaveBeenCalledWith(
+        expect.not.objectContaining({
+          metadata: expect.any(Object),
+        }),
+      );
+    });
+
+    it('should NOT include metadata for other auth types', async () => {
+      const otherAuthConfig = {
+        getContentGeneratorConfig: vi.fn().mockReturnValue({
+          authType: 'gemini-api-key',
+          enableOpenAILogging: false,
+        }),
+        getSessionId: vi.fn().mockReturnValue('other-session-id'),
+      } as unknown as Config;
+
+      const otherGenerator = new OpenAIContentGenerator(
+        'test-key',
+        'gpt-4',
+        otherAuthConfig,
+      );
+
+      const mockResponse = {
+        id: 'chatcmpl-123',
+        choices: [
+          {
+            index: 0,
+            message: { role: 'assistant', content: 'Response' },
+            finish_reason: 'stop',
+          },
+        ],
+        created: 1677652288,
+        model: 'gpt-4',
+      };
+
+      mockOpenAIClient.chat.completions.create.mockResolvedValue(mockResponse);
+
+      const request: GenerateContentParameters = {
+        contents: [{ role: 'user', parts: [{ text: 'Hello' }] }],
+        model: 'gpt-4',
+      };
+
+      await otherGenerator.generateContent(request, 'other-prompt-id');
+
+      // Should NOT include metadata
+      expect(mockOpenAIClient.chat.completions.create).toHaveBeenCalledWith(
+        expect.not.objectContaining({
+          metadata: expect.any(Object),
+        }),
+      );
+    });
+
+    it('should NOT include metadata for other base URLs', async () => {
+      // Mock environment to set a different base URL
+      vi.stubEnv('OPENAI_BASE_URL', 'https://api.openai.com/v1');
+
+      const otherBaseUrlConfig = {
+        getContentGeneratorConfig: vi.fn().mockReturnValue({
+          authType: 'openai',
+          enableOpenAILogging: false,
+        }),
+        getSessionId: vi.fn().mockReturnValue('other-base-url-session-id'),
+      } as unknown as Config;
+
+      const otherBaseUrlGenerator = new OpenAIContentGenerator(
+        'test-key',
+        'gpt-4',
+        otherBaseUrlConfig,
+      );
+
+      const mockResponse = {
+        id: 'chatcmpl-123',
+        choices: [
+          {
+            index: 0,
+            message: { role: 'assistant', content: 'Response' },
+            finish_reason: 'stop',
+          },
+        ],
+        created: 1677652288,
+        model: 'gpt-4',
+      };
+
+      mockOpenAIClient.chat.completions.create.mockResolvedValue(mockResponse);
+
+      const request: GenerateContentParameters = {
+        contents: [{ role: 'user', parts: [{ text: 'Hello' }] }],
+        model: 'gpt-4',
+      };
+
+      await otherBaseUrlGenerator.generateContent(
+        request,
+        'other-base-url-prompt-id',
+      );
+
+      // Should NOT include metadata
+      expect(mockOpenAIClient.chat.completions.create).toHaveBeenCalledWith(
+        expect.not.objectContaining({
+          metadata: expect.any(Object),
+        }),
+      );
+    });
+
+    it('should include metadata in streaming requests when conditions are met', async () => {
+      const qwenConfig = {
+        getContentGeneratorConfig: vi.fn().mockReturnValue({
+          authType: 'qwen-oauth',
+          enableOpenAILogging: false,
+        }),
+        getSessionId: vi.fn().mockReturnValue('streaming-session-id'),
+      } as unknown as Config;
+
+      const qwenGenerator = new OpenAIContentGenerator(
+        'test-key',
+        'qwen-turbo',
+        qwenConfig,
+      );
+
+      const mockStream = [
+        {
+          id: 'chatcmpl-123',
+          choices: [
+            {
+              index: 0,
+              delta: { content: 'Hello' },
+              finish_reason: null,
+            },
+          ],
+          created: 1677652288,
+        },
+        {
+          id: 'chatcmpl-123',
+          choices: [
+            {
+              index: 0,
+              delta: { content: ' there!' },
+              finish_reason: 'stop',
+            },
+          ],
+          created: 1677652288,
+        },
+      ];
+
+      mockOpenAIClient.chat.completions.create.mockResolvedValue({
+        async *[Symbol.asyncIterator]() {
+          for (const chunk of mockStream) {
+            yield chunk;
+          }
+        },
+      });
+
+      const request: GenerateContentParameters = {
+        contents: [{ role: 'user', parts: [{ text: 'Hello' }] }],
+        model: 'qwen-turbo',
+      };
+
+      const stream = await qwenGenerator.generateContentStream(
+        request,
+        'streaming-prompt-id',
+      );
+
+      // Verify metadata was included in the streaming request
+      expect(mockOpenAIClient.chat.completions.create).toHaveBeenCalledWith(
+        expect.objectContaining({
+          metadata: {
+            sessionId: 'streaming-session-id',
+            promptId: 'streaming-prompt-id',
+          },
+          stream: true,
+          stream_options: { include_usage: true },
+        }),
+      );
+
+      // Consume the stream to complete the test
+      const responses = [];
+      for await (const response of stream) {
+        responses.push(response);
+      }
+      expect(responses).toHaveLength(2);
+    });
+
+    it('should NOT include metadata in streaming requests when conditions are not met', async () => {
+      const regularConfig = {
+        getContentGeneratorConfig: vi.fn().mockReturnValue({
+          authType: 'openai',
+          enableOpenAILogging: false,
+        }),
+        getSessionId: vi.fn().mockReturnValue('regular-streaming-session-id'),
+      } as unknown as Config;
+
+      const regularGenerator = new OpenAIContentGenerator(
+        'test-key',
+        'gpt-4',
+        regularConfig,
+      );
+
+      const mockStream = [
+        {
+          id: 'chatcmpl-123',
+          choices: [
+            {
+              index: 0,
+              delta: { content: 'Hello' },
+              finish_reason: null,
+            },
+          ],
+          created: 1677652288,
+        },
+        {
+          id: 'chatcmpl-123',
+          choices: [
+            {
+              index: 0,
+              delta: { content: ' there!' },
+              finish_reason: 'stop',
+            },
+          ],
+          created: 1677652288,
+        },
+      ];
+
+      mockOpenAIClient.chat.completions.create.mockResolvedValue({
+        async *[Symbol.asyncIterator]() {
+          for (const chunk of mockStream) {
+            yield chunk;
+          }
+        },
+      });
+
+      const request: GenerateContentParameters = {
+        contents: [{ role: 'user', parts: [{ text: 'Hello' }] }],
+        model: 'gpt-4',
+      };
+
+      const stream = await regularGenerator.generateContentStream(
+        request,
+        'regular-streaming-prompt-id',
+      );
+
+      // Verify metadata was NOT included in the streaming request
+      expect(mockOpenAIClient.chat.completions.create).toHaveBeenCalledWith(
+        expect.not.objectContaining({
+          metadata: expect.any(Object),
+        }),
+      );
+
+      // Consume the stream to complete the test
+      const responses = [];
+      for await (const response of stream) {
+        responses.push(response);
+      }
+      expect(responses).toHaveLength(2);
+    });
+
+    it('should handle undefined sessionId gracefully', async () => {
+      const qwenConfig = {
+        getContentGeneratorConfig: vi.fn().mockReturnValue({
+          authType: 'qwen-oauth',
+          enableOpenAILogging: false,
+        }),
+        getSessionId: vi.fn().mockReturnValue(undefined), // Undefined session ID
+      } as unknown as Config;
+
+      const qwenGenerator = new OpenAIContentGenerator(
+        'test-key',
+        'qwen-turbo',
+        qwenConfig,
+      );
+
+      const mockResponse = {
+        id: 'chatcmpl-123',
+        choices: [
+          {
+            index: 0,
+            message: { role: 'assistant', content: 'Response' },
+            finish_reason: 'stop',
+          },
+        ],
+        created: 1677652288,
+        model: 'qwen-turbo',
+      };
+
+      mockOpenAIClient.chat.completions.create.mockResolvedValue(mockResponse);
+
+      const request: GenerateContentParameters = {
+        contents: [{ role: 'user', parts: [{ text: 'Hello' }] }],
+        model: 'qwen-turbo',
+      };
+
+      await qwenGenerator.generateContent(
+        request,
+        'undefined-session-prompt-id',
+      );
+
+      expect(mockOpenAIClient.chat.completions.create).toHaveBeenCalledWith(
+        expect.objectContaining({
+          metadata: {
+            sessionId: undefined,
+            promptId: 'undefined-session-prompt-id',
+          },
+        }),
+      );
+    });
+
+    it('should handle undefined baseURL gracefully', async () => {
+      // Ensure no base URL is set
+      vi.stubEnv('OPENAI_BASE_URL', '');
+
+      const noBaseUrlConfig = {
+        getContentGeneratorConfig: vi.fn().mockReturnValue({
+          authType: 'openai',
+          enableOpenAILogging: false,
+        }),
+        getSessionId: vi.fn().mockReturnValue('no-base-url-session-id'),
+      } as unknown as Config;
+
+      const noBaseUrlGenerator = new OpenAIContentGenerator(
+        'test-key',
+        'gpt-4',
+        noBaseUrlConfig,
+      );
+
+      const mockResponse = {
+        id: 'chatcmpl-123',
+        choices: [
+          {
+            index: 0,
+            message: { role: 'assistant', content: 'Response' },
+            finish_reason: 'stop',
+          },
+        ],
+        created: 1677652288,
+        model: 'gpt-4',
+      };
+
+      mockOpenAIClient.chat.completions.create.mockResolvedValue(mockResponse);
+
+      const request: GenerateContentParameters = {
+        contents: [{ role: 'user', parts: [{ text: 'Hello' }] }],
+        model: 'gpt-4',
+      };
+
+      await noBaseUrlGenerator.generateContent(
+        request,
+        'no-base-url-prompt-id',
+      );
+
+      // Should NOT include metadata when baseURL is empty
+      expect(mockOpenAIClient.chat.completions.create).toHaveBeenCalledWith(
+        expect.not.objectContaining({
+          metadata: expect.any(Object),
+        }),
+      );
+    });
+
+    it('should handle undefined authType gracefully', async () => {
+      const undefinedAuthConfig = {
+        getContentGeneratorConfig: vi.fn().mockReturnValue({
+          authType: undefined, // Undefined auth type
+          enableOpenAILogging: false,
+        }),
+        getSessionId: vi.fn().mockReturnValue('undefined-auth-session-id'),
+      } as unknown as Config;
+
+      const undefinedAuthGenerator = new OpenAIContentGenerator(
+        'test-key',
+        'gpt-4',
+        undefinedAuthConfig,
+      );
+
+      const mockResponse = {
+        id: 'chatcmpl-123',
+        choices: [
+          {
+            index: 0,
+            message: { role: 'assistant', content: 'Response' },
+            finish_reason: 'stop',
+          },
+        ],
+        created: 1677652288,
+        model: 'gpt-4',
+      };
+
+      mockOpenAIClient.chat.completions.create.mockResolvedValue(mockResponse);
+
+      const request: GenerateContentParameters = {
+        contents: [{ role: 'user', parts: [{ text: 'Hello' }] }],
+        model: 'gpt-4',
+      };
+
+      await undefinedAuthGenerator.generateContent(
+        request,
+        'undefined-auth-prompt-id',
+      );
+
+      // Should NOT include metadata when authType is undefined
+      expect(mockOpenAIClient.chat.completions.create).toHaveBeenCalledWith(
+        expect.not.objectContaining({
+          metadata: expect.any(Object),
+        }),
+      );
+    });
+
+    it('should handle undefined config gracefully', async () => {
+      const undefinedConfig = {
+        getContentGeneratorConfig: vi.fn().mockReturnValue(undefined), // Undefined config
+        getSessionId: vi.fn().mockReturnValue('undefined-config-session-id'),
+      } as unknown as Config;
+
+      const undefinedConfigGenerator = new OpenAIContentGenerator(
+        'test-key',
+        'gpt-4',
+        undefinedConfig,
+      );
+
+      const mockResponse = {
+        id: 'chatcmpl-123',
+        choices: [
+          {
+            index: 0,
+            message: { role: 'assistant', content: 'Response' },
+            finish_reason: 'stop',
+          },
+        ],
+        created: 1677652288,
+        model: 'gpt-4',
+      };
+
+      mockOpenAIClient.chat.completions.create.mockResolvedValue(mockResponse);
+
+      const request: GenerateContentParameters = {
+        contents: [{ role: 'user', parts: [{ text: 'Hello' }] }],
+        model: 'gpt-4',
+      };
+
+      await undefinedConfigGenerator.generateContent(
+        request,
+        'undefined-config-prompt-id',
+      );
+
+      // Should NOT include metadata when config is undefined
+      expect(mockOpenAIClient.chat.completions.create).toHaveBeenCalledWith(
+        expect.not.objectContaining({
+          metadata: expect.any(Object),
+        }),
+      );
+    });
+  });
 });
diff --git a/packages/core/src/core/openaiContentGenerator.ts b/packages/core/src/core/openaiContentGenerator.ts
index 4e9e73da..311fa6f5 100644
--- a/packages/core/src/core/openaiContentGenerator.ts
+++ b/packages/core/src/core/openaiContentGenerator.ts
@@ -20,7 +20,7 @@ import {
   FunctionCall,
   FunctionResponse,
 } from '@google/genai';
-import { ContentGenerator } from './contentGenerator.js';
+import { AuthType, ContentGenerator } from './contentGenerator.js';
 import OpenAI from 'openai';
 import { logApiResponse } from '../telemetry/loggers.js';
 import { ApiResponseEvent } from '../telemetry/types.js';
@@ -185,6 +185,46 @@ export class OpenAIContentGenerator implements ContentGenerator {
     );
   }
 
+  /**
+   * Determine if metadata should be included in the request.
+   * Only include the `metadata` field if the provider is QWEN_OAUTH
+   * or the baseUrl is 'https://dashscope.aliyuncs.com/compatible-mode/v1'.
+   * This is because some models/providers do not support metadata or need extra configuration.
+   *
+   * @returns true if metadata should be included, false otherwise
+   */
+  private shouldIncludeMetadata(): boolean {
+    const authType = this.config.getContentGeneratorConfig?.()?.authType;
+    // baseUrl may be undefined; default to empty string if so
+    const baseUrl = this.client?.baseURL || '';
+
+    return (
+      authType === AuthType.QWEN_OAUTH ||
+      baseUrl === 'https://dashscope.aliyuncs.com/compatible-mode/v1'
+    );
+  }
+
+  /**
+   * Build metadata object for OpenAI API requests.
+   *
+   * @param userPromptId The user prompt ID to include in metadata
+   * @returns metadata object if shouldIncludeMetadata() returns true, undefined otherwise
+   */
+  private buildMetadata(
+    userPromptId: string,
+  ): { metadata: { sessionId?: string; promptId: string } } | undefined {
+    if (!this.shouldIncludeMetadata()) {
+      return undefined;
+    }
+
+    return {
+      metadata: {
+        sessionId: this.config.getSessionId?.(),
+        promptId: userPromptId,
+      },
+    };
+  }
+
   async generateContent(
     request: GenerateContentParameters,
     userPromptId: string,
@@ -205,10 +245,7 @@ export class OpenAIContentGenerator implements ContentGenerator {
         model: this.model,
         messages,
         ...samplingParams,
-        metadata: {
-          sessionId: this.config.getSessionId?.(),
-          promptId: userPromptId,
-        },
+        ...(this.buildMetadata(userPromptId) || {}),
       };
 
       if (request.config?.tools) {
@@ -339,10 +376,7 @@ export class OpenAIContentGenerator implements ContentGenerator {
         ...samplingParams,
         stream: true,
         stream_options: { include_usage: true },
-        metadata: {
-          sessionId: this.config.getSessionId?.(),
-          promptId: userPromptId,
-        },
+        ...(this.buildMetadata(userPromptId) || {}),
       };
 
       if (request.config?.tools) {