[BUG] Dashscope provider fails with "[] is too short - 'tools'" error when using Deep Research

[xdien]

Description

When using Dashscope (Alibaba Cloud) as a custom LLM provider, the application throws a litellm.BadRequestError with message "DashscopeException - [] is too short - 'tools'" when attempting to use Deep Research functionality.

Steps to Reproduce

1. Install Onyx using the quick start script:

   curl -fsSL https://onyx.app/install_onyx.sh | bash

   Select "lite" option during installation.

2. Navigate to Admin Panel → Language Models → Custom Models

3. Configure the custom model with the following settings:

   • Provider: dashscope
   • Base URL: https://coding-intl.dashscope.aliyuncs.com/v1

4. Click Save and exit Admin Panel

5. Create a New Session

6. Select Deep Research mode

7. Enter any research prompt

8. Expected: The model should process the research request
   Actual: Error occurs: litellm.BadRequestError: DashscopeException - [] is too short - 'tools'

Error Message

litellm.BadRequestError: DashscopeException - [] is too short - 'tools'
Traceback (most recent call last):
  File "/usr/local/lib/python3.11/site-packages/litellm/llms/openai/openai.py", line 762, in completion
    raise e
  File "/usr/local/lib/python3.11/site-packages/litellm/llms/openai/openai.py", line 645, in completion
    return self.streaming(
           ^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.11/site-packages/litellm/llms/openai/openai.py", line 935, in streaming
    headers, response = self.make_sync_openai_chat_completion_request(
                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.11/site-packages/litellm/litellm_core_utils/logging_utils.py", line 237, in sync_wrapper
    result = func(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.11/site-packages/litellm/llms/openai/openai.py", line 502, in make_sync_openai_chat_completion_request
    raise e
  File "/usr/local/lib/python3.11/site-packages/litellm/llms/openai/openai.py", line 477, in make_sync_openai_chat_completion_request
    raw_response = openai_client.chat.completions.with_raw_response.create(
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.11/site-packages/openai/_legacy_response.py", line 364, in wrapped
    return cast(LegacyAPIResponse[R], func(*args, **kwargs))
                                      ^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.11/site-packages/openai/_utils/_utils.py", line 286, in wrapper
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.11/site-packages/openai/resources/chat/completions/completions.py", line 1192, in create
    return self._post(
           ^^^^^^^^^^^
  File "/usr/local/lib/python3.11/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 "/usr/local/lib/python3.11/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': {'code': None, 'message': "[] is too short - 'tools'", 'param': None, 'type': 'invalid_request_error'}, 'request_id': '2085ac25-b6b7-960e-9e75-9a97c67752eb'}

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/usr/local/lib/python3.11/site-packages/litellm/main.py", line 2531, in completion
    raise e
  File "/usr/local/lib/python3.11/site-packages/litellm/main.py", line 2503, in completion
    response = openai_chat_completions.completion(
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.11/site-packages/litellm/llms/openai/openai.py", line 773, in completion
    raise OpenAIError(
litellm.llms.openai.common_utils.OpenAIError: Error code: 400 - {'error': {'code': None, 'message': "[] is too short - 'tools'", 'param': None, 'type': 'invalid_request_error'}, 'request_id': '2085ac25-b6b7-960e-9e75-9a97c67752eb'}

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "/app/onyx/chat/process_message.py", line 1166, in _run_model
    run_deep_research_llm_loop(
  File "/app/onyx/utils/timing.py", line 52, in wrapped_func
    result = func(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^
  File "/app/onyx/deep_research/dr_loop.py", line 351, in run_deep_research_llm_loop
    packet = next(research_plan_generator)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/app/onyx/chat/llm_step.py", line 1148, in run_llm_step_pkt_generator
    for packet in llm.stream(
  File "/app/onyx/llm/multi_llm.py", line 843, in stream
    self._completion(
  File "/app/onyx/llm/multi_llm.py", line 679, in _completion
    raise e
  File "/app/onyx/llm/multi_llm.py", line 654, in _completion
    response = litellm.completion(
               ^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.11/site-packages/litellm/utils.py", line 1742, in wrapper
    raise e
  File "/usr/local/lib/python3.11/site-packages/litellm/utils.py", line 1563, in wrapper
    result = original_function(*args, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.11/site-packages/litellm/main.py", line 4242, in completion
    raise exception_type(
          ^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.11/site-packages/litellm/litellm_core_utils/exception_mapping_utils.py", line 2378, in exception_type
    raise e
  File "/usr/local/lib/python3.11/site-packages/litellm/litellm_core_utils/exception_mapping_utils.py", line 444, in exception_type
    raise BadRequestError(
litellm.exceptions.BadRequestError: litellm.BadRequestError: DashscopeException - [] is too short - 'tools'

Environment

• Installation Method: Docker (via install script)
• Deployment Type: Lite
• LLM Provider: Dashscope (Alibaba Cloud)
• Base URL: https://coding-intl.dashscope.aliyuncs.com/v1

Additional Context

The error suggests that the tools parameter being sent to the Dashscope API is an empty array [], which Dashscope rejects as being "too short". This likely occurs because Deep Research mode may be sending tool definitions that are either empty or not properly formatted for the Dashscope API.