how to use qwen

[idiomer2]

BASE_URL=https://dashscope.aliyuncs.com/compatible-mode/v1

When i change model from gpt-4o to qwen-plus, ChatDev 2.0 backend raise error below. How can i fix it? please give some help

Traceback (most recent call last):
File "F:\code\github\ChatDev\runtime\node\executor\agent_executor.py", line 112, in execute
response_obj = self._invoke_provider(
^^^^^^^^^^^^^^^^^^^^^^
File "F:\code\github\ChatDev\runtime\node\executor\agent_executor.py", line 284, in invoke_provider
response = self.execute_with_retry(node, retry_policy, call_provider)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\code\github\ChatDev\runtime\node\executor\agent_executor.py", line 347, in execute_with_retry
return retrier(func)
^^^^^^^^^^^^^
File "F:\code\github\ChatDev.venv\Lib\site-packages\tenacity_init.py", line 477, in call
do = self.iter(retry_state=retry_state)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\code\github\ChatDev.venv\Lib\site-packages\tenacity_init.py", line 378, in iter
result = action(retry_state)
^^^^^^^^^^^^^^^^^^^
File "F:\code\github\ChatDev.venv\Lib\site-packages\tenacity_init.py", line 420, in exc_check
raise retry_exc.reraise()
^^^^^^^^^^^^^^^^^^^
File "F:\code\github\ChatDev.venv\Lib\site-packages\tenacity_init.py", line 187, in reraise
raise self.last_attempt.result()
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\ProgramData\uv\python\cpython-3.12.12-windows-x86_64-none\Lib\concurrent\futures_base.py", line 449, in result
return self.__get_result()
^^^^^^^^^^^^^^^^^^^
File "D:\ProgramData\uv\python\cpython-3.12.12-windows-x86_64-none\Lib\concurrent\futures_base.py", line 401, in __get_result
raise self.exception
File "F:\code\github\ChatDev.venv\Lib\site-packages\tenacity_init.py", line 480, in call
result = fn(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^
File "F:\code\github\ChatDev\runtime\node\executor\agent_executor.py", line 274, in _call_provider
return provider.call_model(
^^^^^^^^^^^^^^^^^^^^
File "F:\code\github\ChatDev\runtime\node\agent\providers\openai_provider.py", line 76, in call_model
response = client.responses.create(**request_payload)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\code\github\ChatDev.venv\Lib\site-packages\openai\resources\responses\responses.py", line 866, in create
return self._post(
^^^^^^^^^^^
File "F:\code\github\ChatDev.venv\Lib\site-packages\openai_base_client.py", line 1259, in post
return cast(ResponseT, self.request(cast_to, opts, stream=stream, stream_cls=stream_cls))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\code\github\ChatDev.venv\Lib\site-packages\openai_base_client.py", line 1047, in request
raise self._make_status_error_from_response(err.response) from None
openai.NotFoundError: Error code: 404
ERROR:session_33bacb36-7b85-4488-bce4-348f5be95253:{"timestamp": "2026-01-08 10:08:29.250215+00:00", "log_type": "error", "level": "ERROR", "logger": "session_33bacb36-7b85-4488-bce4-348f5be95253", "message": "[Node: Data Analyst] Error calling model: Error code: 404", "correlation_id": null, "workflow_id": "session_33bacb36-7b85-4488-bce4-348f5be95253", "node_id": null, "event_type": null, "execution_path": ["Data Analyst"], "duration": null}

[NA-Wen]

This looks like a model interface compatibility issue. Currently, ChatDev 2.0 relies on an OpenAI-compatible interface, and there are still some inconsistencies when using certain models (e.g., qwen-plus via DashScope).

As a quick workaround, you may try switching the model provider to gemini first to verify whether the overall workflow can run correctly. And we are actively working on improving compatibility across different model providers and will continue refining support for Qwen and other models. Thanks for reporting this issue and for your patience.

[zxrys]

We have now added logic for the OpenAI Provider to automatically fallback to the chat.completions API. Our tests show that it can successfully request the Qwen model. Please pull the latest code and try again.

[idiomer2]

Great! qwen-plus works!

But now I've encountered a new problem:

I'm using workflow data_visualization_basic.yaml . When Data Analyst using tools, the below error was raised (I have confirmed that qwen-plus support function-calling. here is offical doc)

========================= Error ==================================

Traceback (most recent call last):
File "F:\code\github\ChatDev\runtime\node\executor\agent_executor.py", line 112, in execute
response_obj = self._invoke_provider(
^^^^^^^^^^^^^^^^^^^^^^
File "F:\code\github\ChatDev\runtime\node\executor\agent_executor.py", line 284, in invoke_provider
response = self.execute_with_retry(node, retry_policy, call_provider)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\code\github\ChatDev\runtime\node\executor\agent_executor.py", line 346, in execute_with_retry
return retrier(func)
^^^^^^^^^^^^^
File "F:\code\github\ChatDev.venv\Lib\site-packages\tenacity_init.py", line 477, in call
do = self.iter(retry_state=retry_state)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\code\github\ChatDev.venv\Lib\site-packages\tenacity_init.py", line 378, in iter
result = action(retry_state)
^^^^^^^^^^^^^^^^^^^
File "F:\code\github\ChatDev.venv\Lib\site-packages\tenacity_init.py", line 420, in exc_check
raise retry_exc.reraise()
^^^^^^^^^^^^^^^^^^^
File "F:\code\github\ChatDev.venv\Lib\site-packages\tenacity_init.py", line 187, in reraise
raise self.last_attempt.result()
^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\ProgramData\uv\python\cpython-3.12.12-windows-x86_64-none\Lib\concurrent\futures_base.py", line 449, in result
return self.__get_result()
^^^^^^^^^^^^^^^^^^^
File "D:\ProgramData\uv\python\cpython-3.12.12-windows-x86_64-none\Lib\concurrent\futures_base.py", line 401, in __get_result
raise self.exception
File "F:\code\github\ChatDev.venv\Lib\site-packages\tenacity_init.py", line 480, in call
result = fn(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^
File "F:\code\github\ChatDev\runtime\node\executor\agent_executor.py", line 274, in _call_provider
return provider.call_model(
^^^^^^^^^^^^^^^^^^^^
File "F:\code\github\ChatDev\runtime\node\agent\providers\openai_provider.py", line 84, in call_model
response = client.chat.completions.create(**new_request_payload)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\code\github\ChatDev.venv\Lib\site-packages\openai_utils_utils.py", line 286, in wrapper
return func(*args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^
File "F:\code\github\ChatDev.venv\Lib\site-packages\openai\resources\chat\completions\completions.py", line 1192, in create
return self._post(
^^^^^^^^^^^
File "F:\code\github\ChatDev.venv\Lib\site-packages\openai_base_client.py", line 1259, in post
return cast(ResponseT, self.request(cast_to, opts, stream=stream, stream_cls=stream_cls))
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "F:\code\github\ChatDev.venv\Lib\site-packages\openai_base_client.py", line 1047, in request
raise self._make_status_error_from_response(err.response) from None
openai.BadRequestError: Error code: 400 - {'error': {'message': '<400> InternalError.Algo.InvalidParameter: An assistant message with "tool_calls" must be followed by tool messages responding to each "tool_call_id". The following tool_call_ids did not have response messages: message[3].role', 'type': 'invalid_request_error', 'param': None, 'code': 'invalid_parameter_error'}, 'id': 'chatcmpl-ba7c6e87-e831-9684-a4a7-5bc62d709b00', 'request_id': 'ba7c6e87-e831-9684-a4a7-5bc62d709b00'}
ERROR:session_5182a085-bae2-4f2e-bed4-beebf56bb3bd:{"timestamp": "2026-01-09 01:58:53.374619+00:00", "log_type": "error", "level": "ERROR", "logger": "session_5182a085-bae2-4f2e-bed4-beebf56bb3bd", "message": "[Node: Data Analyst] Error calling model: Error code: 400 - {'error': {'message': '<400> InternalError.Algo.InvalidParameter: An assistant message with "tool_calls" must be followed by tool messages responding to each "tool_call_id". The following tool_call_ids did not have response messages: message[3].role', 'type': 'invalid_request_error', 'param': None, 'code': 'invalid_parameter_error'}, 'id': 'chatcmpl-ba7c6e87-e831-9684-a4a7-5bc62d709b00', 'request_id': 'ba7c6e87-e831-9684-a4a7-5bc62d709b00'}", "correlation_id": null, "workflow_id": "session_5182a085-bae2-4f2e-bed4-beebf56bb3bd", "node_id": null, "event_type": null, "execution_path": ["Data Analyst"], "duration": null}

[zxrys]

Thank you for your feedback! We have fixed the function calling error. Please pull the latest code and try again. However, it's important to note that Qwen currently only supports OpenAI's chat.completions.create interface. This interface cannot pass multimodal data in the return value of function calling, which may prevent some workflows from fully utilizing their capabilities (such as data visualization and 3D modeling workflows, both of which rely on returned image data for decision-making). We are considering adding support for Qwen's native API.