diff --git a/tests/providers/test_openai_provider.py b/tests/providers/test_openai_provider.py
deleted file mode 100644
index 8e86912b..00000000
--- a/tests/providers/test_openai_provider.py
+++ /dev/null
@@ -1,126 +0,0 @@
-import inspect
-from asyncio import iscoroutinefunction
-from unittest.mock import AsyncMock, patch
-
-import pytest
-
-from minion.configs.config import LLMConfig
-from minion.message_types import Message
-from minion.providers.cost import CostManager
-from minion.providers.openai_provider import OpenAIProvider
-
-
-@pytest.fixture
-def mock_openai_client():
-    with patch("openai.AsyncOpenAI") as mock_client:
-        mock_instance = mock_client.return_value
-        mock_instance.chat.completions.create = AsyncMock()
-        yield mock_instance
-
-
-@pytest.fixture
-def openai_provider(mock_openai_client):
-    config = LLMConfig(
-        provider="openai", model="gpt-3.5-turbo", api_key="test_key", base_url="https://api.openai.com/v1"
-    )
-    provider = OpenAIProvider(config)
-    provider.client = mock_openai_client
-    return provider
-
-
-@pytest.mark.asyncio
-async def test_openai_provider_cost_calculation(openai_provider, mock_openai_client):
-    # 模拟 OpenAI API 响应
-    mock_response = AsyncMock()
-    mock_response.choices = [AsyncMock(message=AsyncMock(content="Test response"))]
-    mock_response.usage = AsyncMock(completion_tokens=10)
-    mock_openai_client.chat.completions.create.return_value = mock_response
-
-    # 模拟 CostManager.calculate 方法
-    with patch("minion.providers.cost.CostManager.calculate", return_value=(20, 10)):
-        # 模拟消息
-        messages = [Message(role="user", content="Hello, AI!")]
-
-        # 调用 generate 方法
-        response = await openai_provider.generate(messages)
-
-        # 验证响应内容
-        assert response == "Test response"
-
-        # 验证成本计算
-        cost_manager = openai_provider.get_cost()
-        assert isinstance(cost_manager, CostManager)
-        assert cost_manager.total_prompt_tokens == 20
-        assert cost_manager.total_completion_tokens == 10
-
-        # 验证 OpenAI API 被正确调用
-        mock_openai_client.chat.completions.create.assert_called_once()
-
-
-@pytest.mark.asyncio
-async def test_openai_provider_stream_cost_calculation(openai_provider, mock_openai_client):
-    # Create mock chunks as a list of AsyncMock objects
-    mock_chunks = [
-        AsyncMock(
-            choices=[
-                AsyncMock(
-                    delta=AsyncMock(content="Test "),
-                    finish_reason=None
-                )
-            ],
-            usage=None
-        ),
-        AsyncMock(
-            choices=[
-                AsyncMock(
-                    delta=AsyncMock(content="response "),
-                    finish_reason=None
-                )
-            ],
-            usage=None
-        ),
-        AsyncMock(
-            choices=[
-                AsyncMock(
-                    delta=AsyncMock(content="stream."),
-                    finish_reason="stop"
-                )
-            ],
-            usage={
-                "prompt_tokens": 15,
-                "completion_tokens": 3,
-                "total_tokens": 18
-            }
-        ),
-    ]
-
-    # Create an async generator function that will yield our mock chunks
-    async def mock_stream():
-        for chunk in mock_chunks:
-            yield chunk
-
-    # Create the mock response stream
-    mock_response = mock_stream()
-
-    # Set up the mock client to return our stream
-    mock_openai_client.chat.completions.create.return_value = mock_response
-
-    # 模拟 CostManager.calculate 方法
-    with patch("minion.providers.cost.CostManager.calculate", return_value=(15, 3)):
-        # 模拟消息
-        messages = [Message(role="user", content="Hello, AI!")]
-
-        # 调用 generate_stream 方法并收集结果
-        response = await openai_provider.generate_stream(messages)
-
-        # 验证响应内容
-        assert response == "Test response stream."
-
-        # 验证成本计算
-        cost_manager = openai_provider.get_cost()
-        assert isinstance(cost_manager, CostManager)
-        assert cost_manager.total_prompt_tokens == 15
-        assert cost_manager.total_completion_tokens == 3
-
-        # 验证 OpenAI API 被正确调用
-        mock_openai_client.chat.completions.create.assert_called_once()