How to add params in api_url

[moluzhui]

Self-Check List

Before submitting an issue, please ensure you have completed the following steps:

• I have carefully read the relevant user documentation
• I have reviewed the Frequently Asked Questions
• I have searched and reviewed existing issues to confirm this is not a duplicate problem

Problem Description

Please briefly describe the problem you encountered.

The third-party VLM API carries params, which can be accessed via curl. However, params cannot be configured through evalscope code, making evaluation impossible

EvalScope Version (Required)

v1.0.2

Tools Used

• Native / Native framework
• Opencompass backend
• VLMEvalKit backend
• RAGEval backend
• Perf / Model inference stress testing tool
• Arena / Arena mode

Executed Code or Instructions

Please provide the main code or instructions you executed.

from evalscope import TaskConfig, run_task
from evalscope.constants import EvalType

task_cfg = TaskConfig(
    work_dir='outputs',
    model='gpt-5',
    api_url='https://xxx.xxx.xxx.xxx/openai/deployments/gpt-5?api-version=xxxx-preview',
    api_key='apikey',
    eval_type=EvalType.SERVICE,
    datasets=[
        'real_world_qa',
    ],
    eval_batch_size=5,
    generation_config={
        'max_completion_tokens': 10000, 
        'n': 1,  
    },
)

run_task(task_cfg=task_cfg)

Error Log

Please paste the complete error log or console output.

} prediction failed: due to Error code: 404 - {'error': {'code': '404', 'message': 'Resource not found'}}
Traceback:
Traceback (most recent call last):
  File "/data/zhangxinfeng3/evalscope/evalscope/evalscope/evaluator/evaluator.py", line 182, in get_answers
    task_state = future.result()
                 ^^^^^^^^^^^^^^^
  File "/usr/python/lib/python3.11/concurrent/futures/_base.py", line 449, in result

Running Environment

• Operating System: Centos7.9
• Python Version: python3.11

Additional Information

If there is any other relevant information, please provide it here.

I need to add an extra params at the end of the url and try to get the same error through generation_config["extra_body"]

Access the normal curl command

curl https://xxx/openai/deployments/gpt-5/chat/completions?api-version=xxxx-preview \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer xxxx" \
  -d $'{
    "model": "gpt-5",
    "messages": [
        {
            "content": [
                {
                    "image_url": {
                        "url": "images_url"
                    },
                    "type": "image_url"
                },
                {
                    "text": "What does the picture mainly talk about?",
                    "type": "text"
                }
            ],
            "role": "user"
        }
    ]
}'