Question about the evaluation Setting

[206zlccc]

mcp-atlas/services/mcp_eval/mcp_completion/llm.py

Lines 49 to 56 in 867003a

	response = await litellm.acompletion(
	model=model,
	messages=litellm_messages,
	tools=litellm_tools,
	api_key=config.LLM_API_KEY,
	api_base=config.LLM_BASE_URL,
	timeout=config.DEFAULT_TIMEOUT,
	)

Hi, I found that all model is tested without any reasoning_effort. Are all model results on the leaderboard tested with reasoning_effort disabled? Does MCP-Atlas specifically test tool-calling abilities without the reasoning before tool-calling ?