From 4defed7f2e7eaacdf8e130857eddc6018c7bc3f7 Mon Sep 17 00:00:00 2001
From: yuneng <yuneng@berri.ai>
Date: Sun, 20 Sep 2026 08:03:54 +0000
Subject: [PATCH] test: migrate wave 1 phase 8 legacy llm tests to tests/unit

Co-Authored-By: Devin AI <158243242+devin-ai-integration[bot]@users.noreply.github.com>
---
 .../chat/test_hosted_vllm_ssl_verify.py       | 147 ------------------
 .../test_hosted_vllm_embedding_ssl_verify.py  | 135 ----------------
 ..._github_copilot_messages_transformation.py |   7 -
 ...github_copilot_responses_transformation.py | 116 +++++---------
 .../test_gradient_ai_chat_transformation.py   |   0
 .../chat/test_groq_chat_transformation.py     |   2 -
 .../llms/groq/test_groq_cost_calculator.py    |   0
 .../test_hosted_vllm_chat_transformation.py   |  71 +--------
 ...st_hosted_vllm_embedding_transformation.py |   8 +-
 ...t_hosted_vllm_image_edit_transformation.py |   0
 .../responses/test_hosted_vllm_responses.py   |   9 +-
 .../test_hosted_vllm_rerank_transformation.py |   0
 .../test_hosted_vllm_video_transformation.py  |   0
 .../test_huggingface_rerank_transformation.py |  40 +----
 .../test_inception_chat_transformation.py     |  18 +--
 ...est_inception_completion_transformation.py |  18 +--
 .../test_jina_embedding_transformation.py     |   0
 .../chat/test_langflow_chat_transformation.py |  31 +---
 .../litellm_proxy/test_sandbox_executor.py    |  25 +--
 .../litellm_proxy/test_skills_ownership.py    |  73 ++-------
 20 files changed, 84 insertions(+), 616 deletions(-)
 delete mode 100644 tests/test_litellm/llms/hosted_vllm/chat/test_hosted_vllm_ssl_verify.py
 delete mode 100644 tests/test_litellm/llms/hosted_vllm/embedding/test_hosted_vllm_embedding_ssl_verify.py
 rename tests/{test_litellm => unit}/llms/github_copilot/messages/test_github_copilot_messages_transformation.py (98%)
 rename tests/{test_litellm => unit}/llms/github_copilot/responses/test_github_copilot_responses_transformation.py (89%)
 rename tests/{test_litellm => unit}/llms/gradient_ai/chat/test_gradient_ai_chat_transformation.py (100%)
 rename tests/{test_litellm => unit}/llms/groq/chat/test_groq_chat_transformation.py (99%)
 rename tests/{test_litellm => unit}/llms/groq/test_groq_cost_calculator.py (100%)
 rename tests/{test_litellm => unit}/llms/hosted_vllm/chat/test_hosted_vllm_chat_transformation.py (82%)
 rename tests/{test_litellm => unit}/llms/hosted_vllm/embedding/test_hosted_vllm_embedding_transformation.py (97%)
 rename tests/{test_litellm => unit}/llms/hosted_vllm/image_edit/test_hosted_vllm_image_edit_transformation.py (100%)
 rename tests/{test_litellm => unit}/llms/hosted_vllm/responses/test_hosted_vllm_responses.py (96%)
 rename tests/{test_litellm => unit}/llms/hosted_vllm/test_hosted_vllm_rerank_transformation.py (100%)
 rename tests/{test_litellm => unit}/llms/hosted_vllm/videos/test_hosted_vllm_video_transformation.py (100%)
 rename tests/{test_litellm => unit}/llms/huggingface/rerank/test_huggingface_rerank_transformation.py (91%)
 rename tests/{test_litellm => unit}/llms/inception/test_inception_chat_transformation.py (96%)
 rename tests/{test_litellm => unit}/llms/inception/test_inception_completion_transformation.py (95%)
 rename tests/{test_litellm => unit}/llms/jina_ai/embedding/test_jina_embedding_transformation.py (100%)
 rename tests/{test_litellm => unit}/llms/langflow/chat/test_langflow_chat_transformation.py (93%)
 rename tests/{test_litellm => unit}/llms/litellm_proxy/test_sandbox_executor.py (84%)
 rename tests/{test_litellm => unit}/llms/litellm_proxy/test_skills_ownership.py (88%)

diff --git a/tests/test_litellm/llms/hosted_vllm/chat/test_hosted_vllm_ssl_verify.py b/tests/test_litellm/llms/hosted_vllm/chat/test_hosted_vllm_ssl_verify.py
deleted file mode 100644
index 2364468efe1..00000000000
--- a/tests/test_litellm/llms/hosted_vllm/chat/test_hosted_vllm_ssl_verify.py
+++ /dev/null
@@ -1,147 +0,0 @@
-"""
-Test SSL verification for hosted_vllm provider.
-
-This test ensures that the ssl_verify parameter is properly passed through
-to the HTTP client when using the hosted_vllm provider.
-
-Issue: ssl_verify parameter was being ignored because hosted_vllm fell through
-to the OpenAI catch-all path in main.py, which doesn't pass ssl_verify to the HTTP client.
-"""
-
-from unittest.mock import MagicMock, patch
-
-import pytest
-
-
-import litellm
-
-
-class TestHostedVLLMSSLVerify:
-    """Test suite for SSL verification in hosted_vllm provider."""
-
-    @patch("litellm.llms.custom_httpx.llm_http_handler._get_httpx_client")
-    def test_hosted_vllm_ssl_verify_false_sync(self, mock_get_httpx_client):
-        """Test that ssl_verify=False is passed to the HTTP client for sync calls."""
-        # Setup mock client
-        mock_client = MagicMock()
-        mock_response = MagicMock()
-        mock_response.status_code = 200
-        mock_response.headers = {"content-type": "application/json"}
-        mock_response.json.return_value = {
-            "id": "chatcmpl-test",
-            "object": "chat.completion",
-            "created": 1234567890,
-            "model": "test-model",
-            "choices": [
-                {
-                    "index": 0,
-                    "message": {
-                        "role": "assistant",
-                        "content": "Test response",
-                    },
-                    "finish_reason": "stop",
-                }
-            ],
-            "usage": {
-                "prompt_tokens": 10,
-                "completion_tokens": 5,
-                "total_tokens": 15,
-            },
-        }
-        mock_response.text = '{"id": "chatcmpl-test", "object": "chat.completion", "created": 1234567890, "model": "test-model", "choices": [{"index": 0, "message": {"role": "assistant", "content": "Test response"}, "finish_reason": "stop"}], "usage": {"prompt_tokens": 10, "completion_tokens": 5, "total_tokens": 15}}'
-        mock_client.post.return_value = mock_response
-        mock_get_httpx_client.return_value = mock_client
-
-        try:
-            litellm.completion(
-                model="hosted_vllm/test-model",
-                messages=[{"role": "user", "content": "Hello"}],
-                api_base="https://test-vllm.example.com/v1",
-                ssl_verify=False,
-            )
-        except Exception:
-            # Even if the response parsing fails, we just need to verify
-            # that the mock was called with the correct ssl_verify parameter
-            pass
-
-        # Verify _get_httpx_client was called with ssl_verify=False
-        mock_get_httpx_client.assert_called()
-        call_args = mock_get_httpx_client.call_args
-
-        # Check that params contains ssl_verify=False
-        if call_args[0]:
-            # Positional argument
-            params = call_args[0][0]
-        else:
-            # Keyword argument
-            params = call_args[1].get("params", {})
-
-        assert (
-            params.get("ssl_verify") is False
-        ), f"Expected ssl_verify=False in params, got {params}"
-
-    @patch("litellm.llms.custom_httpx.llm_http_handler.get_async_httpx_client")
-    @pytest.mark.asyncio
-    async def test_hosted_vllm_ssl_verify_false_async(
-        self, mock_get_async_httpx_client
-    ):
-        """Test that ssl_verify=False is passed to the HTTP client for async calls."""
-        # Setup mock async client
-        mock_client = MagicMock()
-        mock_response = MagicMock()
-        mock_response.status_code = 200
-        mock_response.headers = {"content-type": "application/json"}
-        mock_response.json.return_value = {
-            "id": "chatcmpl-test",
-            "object": "chat.completion",
-            "created": 1234567890,
-            "model": "test-model",
-            "choices": [
-                {
-                    "index": 0,
-                    "message": {
-                        "role": "assistant",
-                        "content": "Test response",
-                    },
-                    "finish_reason": "stop",
-                }
-            ],
-            "usage": {
-                "prompt_tokens": 10,
-                "completion_tokens": 5,
-                "total_tokens": 15,
-            },
-        }
-        mock_response.text = '{"id": "chatcmpl-test", "object": "chat.completion", "created": 1234567890, "model": "test-model", "choices": [{"index": 0, "message": {"role": "assistant", "content": "Test response"}, "finish_reason": "stop"}], "usage": {"prompt_tokens": 10, "completion_tokens": 5, "total_tokens": 15}}'
-
-        async def mock_post(*args, **kwargs):
-            return mock_response
-
-        mock_client.post = mock_post
-        mock_get_async_httpx_client.return_value = mock_client
-
-        try:
-            await litellm.acompletion(
-                model="hosted_vllm/test-model",
-                messages=[{"role": "user", "content": "Hello"}],
-                api_base="https://test-vllm.example.com/v1",
-                ssl_verify=False,
-            )
-        except Exception:
-            # Even if the response parsing fails, we just need to verify
-            # that the mock was called with the correct ssl_verify parameter
-            pass
-
-        # Verify get_async_httpx_client was called with ssl_verify=False
-        mock_get_async_httpx_client.assert_called()
-        call_kwargs = mock_get_async_httpx_client.call_args[1]
-
-        # Check that params contains ssl_verify=False
-        params = call_kwargs.get("params", {})
-        assert (
-            params.get("ssl_verify") is False
-        ), f"Expected ssl_verify=False in params, got {params}"
-
-
-if __name__ == "__main__":
-    pytest.main([__file__, "-v", "-s"])
diff --git a/tests/test_litellm/llms/hosted_vllm/embedding/test_hosted_vllm_embedding_ssl_verify.py b/tests/test_litellm/llms/hosted_vllm/embedding/test_hosted_vllm_embedding_ssl_verify.py
deleted file mode 100644
index de94da49384..00000000000
--- a/tests/test_litellm/llms/hosted_vllm/embedding/test_hosted_vllm_embedding_ssl_verify.py
+++ /dev/null
@@ -1,135 +0,0 @@
-"""
-Test SSL verification for hosted_vllm provider embeddings.
-
-This test ensures that the ssl_verify parameter is properly passed through
-to the HTTP client when using the hosted_vllm provider for embeddings.
-
-Issue: ssl_verify parameter was being ignored because hosted_vllm fell through
-to the openai_like catch-all path in main.py, which doesn't pass ssl_verify to the HTTP client.
-"""
-
-from unittest.mock import MagicMock, patch
-
-import pytest
-
-
-import litellm
-
-
-class TestHostedVLLMEmbeddingSSLVerify:
-    """Test suite for SSL verification in hosted_vllm provider embeddings."""
-
-    @patch("litellm.llms.custom_httpx.llm_http_handler._get_httpx_client")
-    def test_hosted_vllm_embedding_ssl_verify_false_sync(self, mock_get_httpx_client):
-        """Test that ssl_verify=False is passed to the HTTP client for sync embedding calls."""
-        # Setup mock client
-        mock_client = MagicMock()
-        mock_response = MagicMock()
-        mock_response.status_code = 200
-        mock_response.headers = {"content-type": "application/json"}
-        mock_response.json.return_value = {
-            "object": "list",
-            "data": [
-                {
-                    "object": "embedding",
-                    "index": 0,
-                    "embedding": [0.1, 0.2, 0.3, 0.4, 0.5],
-                }
-            ],
-            "model": "text-embedding-model",
-            "usage": {
-                "prompt_tokens": 5,
-                "total_tokens": 5,
-            },
-        }
-        mock_response.text = '{"object": "list", "data": [{"object": "embedding", "index": 0, "embedding": [0.1, 0.2, 0.3, 0.4, 0.5]}], "model": "text-embedding-model", "usage": {"prompt_tokens": 5, "total_tokens": 5}}'
-        mock_client.post.return_value = mock_response
-        mock_get_httpx_client.return_value = mock_client
-
-        try:
-            litellm.embedding(
-                model="hosted_vllm/text-embedding-model",
-                input=["hello world"],
-                api_base="https://test-vllm.example.com/v1",
-                ssl_verify=False,
-            )
-        except Exception:
-            # Even if the response parsing fails, we just need to verify
-            # that the mock was called with the correct ssl_verify parameter
-            pass
-
-        # Verify _get_httpx_client was called with ssl_verify=False
-        mock_get_httpx_client.assert_called()
-        call_args = mock_get_httpx_client.call_args
-
-        # Check that params contains ssl_verify=False
-        if call_args[0]:
-            # Positional argument
-            params = call_args[0][0]
-        else:
-            # Keyword argument
-            params = call_args[1].get("params", {})
-
-        assert (
-            params.get("ssl_verify") is False
-        ), f"Expected ssl_verify=False in params, got {params}"
-
-    @patch("litellm.llms.custom_httpx.llm_http_handler.get_async_httpx_client")
-    @pytest.mark.asyncio
-    async def test_hosted_vllm_embedding_ssl_verify_false_async(
-        self, mock_get_async_httpx_client
-    ):
-        """Test that ssl_verify=False is passed to the HTTP client for async embedding calls."""
-        # Setup mock async client
-        mock_client = MagicMock()
-        mock_response = MagicMock()
-        mock_response.status_code = 200
-        mock_response.headers = {"content-type": "application/json"}
-        mock_response.json.return_value = {
-            "object": "list",
-            "data": [
-                {
-                    "object": "embedding",
-                    "index": 0,
-                    "embedding": [0.1, 0.2, 0.3, 0.4, 0.5],
-                }
-            ],
-            "model": "text-embedding-model",
-            "usage": {
-                "prompt_tokens": 5,
-                "total_tokens": 5,
-            },
-        }
-        mock_response.text = '{"object": "list", "data": [{"object": "embedding", "index": 0, "embedding": [0.1, 0.2, 0.3, 0.4, 0.5]}], "model": "text-embedding-model", "usage": {"prompt_tokens": 5, "total_tokens": 5}}'
-
-        async def mock_post(*args, **kwargs):
-            return mock_response
-
-        mock_client.post = mock_post
-        mock_get_async_httpx_client.return_value = mock_client
-
-        try:
-            await litellm.aembedding(
-                model="hosted_vllm/text-embedding-model",
-                input=["hello world"],
-                api_base="https://test-vllm.example.com/v1",
-                ssl_verify=False,
-            )
-        except Exception:
-            # Even if the response parsing fails, we just need to verify
-            # that the mock was called with the correct ssl_verify parameter
-            pass
-
-        # Verify get_async_httpx_client was called with ssl_verify=False
-        mock_get_async_httpx_client.assert_called()
-        call_kwargs = mock_get_async_httpx_client.call_args[1]
-
-        # Check that params contains ssl_verify=False
-        params = call_kwargs.get("params", {})
-        assert (
-            params.get("ssl_verify") is False
-        ), f"Expected ssl_verify=False in params, got {params}"
-
-
-if __name__ == "__main__":
-    pytest.main([__file__, "-v", "-s"])
diff --git a/tests/test_litellm/llms/github_copilot/messages/test_github_copilot_messages_transformation.py b/tests/unit/llms/github_copilot/messages/test_github_copilot_messages_transformation.py
similarity index 98%
rename from tests/test_litellm/llms/github_copilot/messages/test_github_copilot_messages_transformation.py
rename to tests/unit/llms/github_copilot/messages/test_github_copilot_messages_transformation.py
index 8039e744f46..9e9760650cf 100644
--- a/tests/test_litellm/llms/github_copilot/messages/test_github_copilot_messages_transformation.py
+++ b/tests/unit/llms/github_copilot/messages/test_github_copilot_messages_transformation.py
@@ -10,13 +10,6 @@ from litellm.llms.github_copilot.messages.transformation import (
 )
 
 
-def test_github_copilot_anthropic_messages_config_init():
-    """Test GithubCopilotAnthropicMessagesConfig initialization."""
-    config = GithubCopilotAnthropicMessagesConfig()
-    assert config is not None
-    assert hasattr(config, "authenticator")
-
-
 def test_github_copilot_anthropic_messages_get_complete_url():
     """get_complete_url builds the /v1/messages URL from the base it is handed.
 
diff --git a/tests/test_litellm/llms/github_copilot/responses/test_github_copilot_responses_transformation.py b/tests/unit/llms/github_copilot/responses/test_github_copilot_responses_transformation.py
similarity index 89%
rename from tests/test_litellm/llms/github_copilot/responses/test_github_copilot_responses_transformation.py
rename to tests/unit/llms/github_copilot/responses/test_github_copilot_responses_transformation.py
index 0174465b0cc..b8380b7adb4 100644
--- a/tests/test_litellm/llms/github_copilot/responses/test_github_copilot_responses_transformation.py
+++ b/tests/unit/llms/github_copilot/responses/test_github_copilot_responses_transformation.py
@@ -26,9 +26,7 @@ def use_local_model_cost_map(monkeypatch: pytest.MonkeyPatch):
     """Pin litellm.model_cost to the bundled local backup so tests don't depend
     on remote catalog fetches (and don't change behavior across remote refreshes)."""
     monkeypatch.setenv("LITELLM_LOCAL_MODEL_COST_MAP", "True")
-    monkeypatch.setattr(
-        litellm, "model_cost", get_model_cost_map(url=litellm.model_cost_map_url)
-    )
+    monkeypatch.setattr(litellm, "model_cost", get_model_cost_map(url=litellm.model_cost_map_url))
     litellm.add_known_models(model_cost_map=litellm.model_cost)
 
 
@@ -44,49 +42,35 @@ class TestGithubCopilotResponsesAPITransformation:
             provider=LlmProviders.GITHUB_COPILOT,
         )
 
-        assert (
-            config is not None
-        ), "Config should not be None for GitHub Copilot provider"
-        assert isinstance(
-            config, GithubCopilotResponsesAPIConfig
-        ), f"Expected GithubCopilotResponsesAPIConfig, got {type(config)}"
-        assert (
-            config.custom_llm_provider == LlmProviders.GITHUB_COPILOT
-        ), "custom_llm_provider should be GITHUB_COPILOT"
+        assert config is not None, "Config should not be None for GitHub Copilot provider"
+        assert isinstance(config, GithubCopilotResponsesAPIConfig), (
+            f"Expected GithubCopilotResponsesAPIConfig, got {type(config)}"
+        )
+        assert config.custom_llm_provider == LlmProviders.GITHUB_COPILOT, "custom_llm_provider should be GITHUB_COPILOT"
 
     @patch("litellm.llms.github_copilot.responses.transformation.Authenticator")
     def test_github_copilot_responses_endpoint_url(self, mock_authenticator_class):
         """Test that get_complete_url returns correct GitHub Copilot endpoint"""
         # Mock authenticator to return default base
         mock_auth_instance = MagicMock()
-        mock_auth_instance.get_api_base.return_value = (
-            "https://api.individual.githubcopilot.com"
-        )
+        mock_auth_instance.get_api_base.return_value = "https://api.individual.githubcopilot.com"
         mock_authenticator_class.return_value = mock_auth_instance
 
         config = GithubCopilotResponsesAPIConfig()
 
         # Test with default GitHub Copilot API base (from authenticator)
         url = config.get_complete_url(api_base=None, litellm_params={})
-        assert (
-            url == "https://api.individual.githubcopilot.com/responses"
-        ), f"Expected GitHub Copilot responses endpoint, got {url}"
+        assert url == "https://api.individual.githubcopilot.com/responses", (
+            f"Expected GitHub Copilot responses endpoint, got {url}"
+        )
 
         # Test with custom api_base (overrides authenticator)
-        custom_url = config.get_complete_url(
-            api_base="https://custom.githubcopilot.com", litellm_params={}
-        )
-        assert (
-            custom_url == "https://custom.githubcopilot.com/responses"
-        ), f"Expected custom endpoint, got {custom_url}"
+        custom_url = config.get_complete_url(api_base="https://custom.githubcopilot.com", litellm_params={})
+        assert custom_url == "https://custom.githubcopilot.com/responses", f"Expected custom endpoint, got {custom_url}"
 
         # Test with trailing slash
-        url_with_slash = config.get_complete_url(
-            api_base="https://api.githubcopilot.com/", litellm_params={}
-        )
-        assert (
-            url_with_slash == "https://api.githubcopilot.com/responses"
-        ), "Should handle trailing slash"
+        url_with_slash = config.get_complete_url(api_base="https://api.githubcopilot.com/", litellm_params={})
+        assert url_with_slash == "https://api.githubcopilot.com/responses", "Should handle trailing slash"
 
     @patch("litellm.llms.github_copilot.responses.transformation.Authenticator")
     def test_validate_environment_default_headers(self, mock_authenticator_class):
@@ -98,9 +82,7 @@ class TestGithubCopilotResponsesAPITransformation:
 
         config = GithubCopilotResponsesAPIConfig()
 
-        headers = config.validate_environment(
-            headers={}, model="gpt-5.1-codex", litellm_params={}
-        )
+        headers = config.validate_environment(headers={}, model="gpt-5.1-codex", litellm_params={})
 
         # Check required headers
         assert headers["Authorization"] == "Bearer test-api-key-123"
@@ -127,9 +109,7 @@ class TestGithubCopilotResponsesAPITransformation:
             "custom-header": "custom-value",
         }
 
-        headers = config.validate_environment(
-            headers=custom_headers, model="gpt-5.1-codex", litellm_params={}
-        )
+        headers = config.validate_environment(headers=custom_headers, model="gpt-5.1-codex", litellm_params={})
 
         # User header should override default
         assert headers["editor-version"] == "custom/2.0.0"
@@ -182,9 +162,7 @@ class TestGithubCopilotResponsesAPITransformation:
         """Test _has_vision_input detects input_image type"""
         config = GithubCopilotResponsesAPIConfig()
 
-        input_with_vision = [
-            {"role": "user", "content": [{"type": "input_image", "data": "base64..."}]}
-        ]
+        input_with_vision = [{"role": "user", "content": [{"type": "input_image", "data": "base64..."}]}]
 
         has_vision = config._has_vision_input(input_with_vision)
         assert has_vision is True, "Should detect input_image type"
@@ -246,13 +224,11 @@ class TestGithubCopilotResponsesAPITransformation:
             }
         ]
 
-        headers = config.validate_environment(
-            headers={}, model="gpt-5.1-codex", litellm_params=mock_litellm_params
-        )
+        headers = config.validate_environment(headers={}, model="gpt-5.1-codex", litellm_params=mock_litellm_params)
 
-        assert (
-            headers.get("copilot-vision-request") == "true"
-        ), "Should add copilot-vision-request header for vision input"
+        assert headers.get("copilot-vision-request") == "true", (
+            "Should add copilot-vision-request header for vision input"
+        )
 
     @patch("litellm.llms.github_copilot.responses.transformation.Authenticator")
     def test_validate_environment_with_x_initiator(self, mock_authenticator_class):
@@ -270,21 +246,15 @@ class TestGithubCopilotResponsesAPITransformation:
             {"role": "assistant", "content": "Hi"},
         ]
 
-        headers = config.validate_environment(
-            headers={}, model="gpt-5.1-codex", litellm_params=mock_litellm_params
-        )
+        headers = config.validate_environment(headers={}, model="gpt-5.1-codex", litellm_params=mock_litellm_params)
 
-        assert (
-            headers.get("X-Initiator") == "agent"
-        ), "Should set X-Initiator to 'agent' for assistant role"
+        assert headers.get("X-Initiator") == "agent", "Should set X-Initiator to 'agent' for assistant role"
 
     def test_map_openai_params_no_transformation(self):
         """Test that map_openai_params passes through parameters unchanged"""
         config = GithubCopilotResponsesAPIConfig()
 
-        params = ResponsesAPIOptionalRequestParams(
-            temperature=0.7, max_output_tokens=1000, stream=False
-        )
+        params = ResponsesAPIOptionalRequestParams(temperature=0.7, max_output_tokens=1000, stream=False)
 
         result = config.map_openai_params(
             response_api_optional_params=params,
@@ -338,9 +308,9 @@ class TestGithubCopilotResponsesAPITransformation:
         result = config._handle_reasoning_item(reasoning_item)
 
         # encrypted_content should be preserved
-        assert (
-            result.get("encrypted_content") == "encrypted-blob-abc123"
-        ), "encrypted_content must be preserved for GitHub Copilot multi-turn conversations"
+        assert result.get("encrypted_content") == "encrypted-blob-abc123", (
+            "encrypted_content must be preserved for GitHub Copilot multi-turn conversations"
+        )
         # status=None should be filtered out
         assert "status" not in result, "status=None should be filtered out"
         # content=None should be filtered out
@@ -393,9 +363,7 @@ class TestGithubCopilotResponsesAPIRouting:
     in the (already-merged) model info; otherwise returns None so the dispatcher
     routes through the chat-completions translation bridge."""
 
-    @patch(
-        "litellm.llms.github_copilot.responses.transformation._cached_get_model_info_helper"
-    )
+    @patch("litellm.llms.github_copilot.responses.transformation._cached_get_model_info_helper")
     def test_returns_config_when_mode_is_responses(self, mock_get_info):
         """``mode=responses`` returns native config."""
         mock_get_info.return_value = {"mode": "responses"}
@@ -405,9 +373,7 @@ class TestGithubCopilotResponsesAPIRouting:
         )
         assert isinstance(config, GithubCopilotResponsesAPIConfig)
 
-    @patch(
-        "litellm.llms.github_copilot.responses.transformation._cached_get_model_info_helper"
-    )
+    @patch("litellm.llms.github_copilot.responses.transformation._cached_get_model_info_helper")
     def test_returns_none_when_mode_is_chat(self, mock_get_info):
         """``mode=chat`` returns None so dispatcher uses bridge."""
         mock_get_info.return_value = {"mode": "chat"}
@@ -417,9 +383,7 @@ class TestGithubCopilotResponsesAPIRouting:
         )
         assert config is None
 
-    @patch(
-        "litellm.llms.github_copilot.responses.transformation._cached_get_model_info_helper"
-    )
+    @patch("litellm.llms.github_copilot.responses.transformation._cached_get_model_info_helper")
     def test_returns_none_when_mode_is_unset_and_no_endpoints(self, mock_get_info):
         """Entry without ``mode`` and without ``supported_endpoints`` returns None
         (conservative default)."""
@@ -499,9 +463,7 @@ class TestGithubCopilotResponsesAPIRouting:
         )
         assert isinstance(config, GithubCopilotResponsesAPIConfig)
 
-    @patch(
-        "litellm.llms.github_copilot.responses.transformation._cached_get_model_info_helper"
-    )
+    @patch("litellm.llms.github_copilot.responses.transformation._cached_get_model_info_helper")
     def test_returns_none_when_get_model_info_raises(self, mock_get_info):
         """Catalog lookup failure (model not registered) returns None
         (conservative default; bridge handles unknown models safely)."""
@@ -512,9 +474,7 @@ class TestGithubCopilotResponsesAPIRouting:
         )
         assert config is None
 
-    @patch(
-        "litellm.llms.github_copilot.responses.transformation._cached_get_model_info_helper"
-    )
+    @patch("litellm.llms.github_copilot.responses.transformation._cached_get_model_info_helper")
     def test_user_override_via_register_model(self, mock_get_info):
         """User-supplied per-deployment ``model_info`` flows through
         ``litellm.register_model`` (called by the router) into the merged
@@ -528,9 +488,7 @@ class TestGithubCopilotResponsesAPIRouting:
         )
         assert isinstance(config, GithubCopilotResponsesAPIConfig)
 
-    @patch(
-        "litellm.llms.github_copilot.responses.transformation._cached_get_model_info_helper"
-    )
+    @patch("litellm.llms.github_copilot.responses.transformation._cached_get_model_info_helper")
     def test_realistic_chat_only_entry_returns_none(self, mock_get_info):
         """Realistic ``model_prices_and_context_window.json`` shape for a
         chat-only Copilot model (e.g. github_copilot/gemini-3.1-pro-preview)
@@ -554,9 +512,7 @@ class TestGithubCopilotResponsesAPIRouting:
         )
         assert config is None
 
-    @patch(
-        "litellm.llms.github_copilot.responses.transformation._cached_get_model_info_helper"
-    )
+    @patch("litellm.llms.github_copilot.responses.transformation._cached_get_model_info_helper")
     def test_realistic_responses_only_entry_returns_config(self, mock_get_info):
         """Realistic catalog entry for a Responses-only Copilot model
         (e.g. github_copilot/gpt-5.5) returns the native config."""
@@ -592,9 +548,7 @@ class TestGithubCopilotReasoningStreamItemIdNormalization:
     output_index group to the id from its output_item.added."""
 
     def _config(self):
-        with patch(
-            "litellm.llms.github_copilot.responses.transformation.Authenticator"
-        ):
+        with patch("litellm.llms.github_copilot.responses.transformation.Authenticator"):
             return GithubCopilotResponsesAPIConfig()
 
     def _transform(self, config, chunk):
diff --git a/tests/test_litellm/llms/gradient_ai/chat/test_gradient_ai_chat_transformation.py b/tests/unit/llms/gradient_ai/chat/test_gradient_ai_chat_transformation.py
similarity index 100%
rename from tests/test_litellm/llms/gradient_ai/chat/test_gradient_ai_chat_transformation.py
rename to tests/unit/llms/gradient_ai/chat/test_gradient_ai_chat_transformation.py
diff --git a/tests/test_litellm/llms/groq/chat/test_groq_chat_transformation.py b/tests/unit/llms/groq/chat/test_groq_chat_transformation.py
similarity index 99%
rename from tests/test_litellm/llms/groq/chat/test_groq_chat_transformation.py
rename to tests/unit/llms/groq/chat/test_groq_chat_transformation.py
index f605958b979..f5a7a920124 100644
--- a/tests/test_litellm/llms/groq/chat/test_groq_chat_transformation.py
+++ b/tests/unit/llms/groq/chat/test_groq_chat_transformation.py
@@ -202,5 +202,3 @@ class TestGroqWebSearchUsageSignal:
         model_response = litellm.ModelResponse()
         GroqChatConfig()._add_web_search_usage(model_response=model_response)
         assert getattr(model_response, "usage", None) is None
-
-
diff --git a/tests/test_litellm/llms/groq/test_groq_cost_calculator.py b/tests/unit/llms/groq/test_groq_cost_calculator.py
similarity index 100%
rename from tests/test_litellm/llms/groq/test_groq_cost_calculator.py
rename to tests/unit/llms/groq/test_groq_cost_calculator.py
diff --git a/tests/test_litellm/llms/hosted_vllm/chat/test_hosted_vllm_chat_transformation.py b/tests/unit/llms/hosted_vllm/chat/test_hosted_vllm_chat_transformation.py
similarity index 82%
rename from tests/test_litellm/llms/hosted_vllm/chat/test_hosted_vllm_chat_transformation.py
rename to tests/unit/llms/hosted_vllm/chat/test_hosted_vllm_chat_transformation.py
index 82b05601a85..1cc6a1457fc 100644
--- a/tests/test_litellm/llms/hosted_vllm/chat/test_hosted_vllm_chat_transformation.py
+++ b/tests/unit/llms/hosted_vllm/chat/test_hosted_vllm_chat_transformation.py
@@ -41,74 +41,9 @@ def test_hosted_vllm_chat_transformation_file_url():
     ]
 
 
-def test_hosted_vllm_chat_transformation_with_audio_url():
-    from litellm import completion
-
-    mock_client = MagicMock()
-    mock_response = MagicMock()
-    mock_response.status_code = 200
-    mock_response.headers = {"content-type": "application/json"}
-    mock_response.json.return_value = {
-        "id": "chatcmpl-test",
-        "object": "chat.completion",
-        "created": 1234567890,
-        "model": "llama-3.1-70b-instruct",
-        "choices": [
-            {
-                "index": 0,
-                "message": {"role": "assistant", "content": "Test response"},
-                "finish_reason": "stop",
-            }
-        ],
-        "usage": {"prompt_tokens": 10, "completion_tokens": 5, "total_tokens": 15},
-    }
-    mock_response.text = json.dumps(mock_response.json.return_value)
-    mock_client.post.return_value = mock_response
-
-    with patch(
-        "litellm.llms.custom_httpx.llm_http_handler._get_httpx_client",
-        return_value=mock_client,
-    ):
-        try:
-            completion(
-                model="hosted_vllm/llama-3.1-70b-instruct",
-                messages=[
-                    {
-                        "role": "user",
-                        "content": [
-                            {
-                                "type": "audio_url",
-                                "audio_url": {"url": "https://example.com/audio.mp3"},
-                            },
-                        ],
-                    },
-                ],
-                api_base="https://test-vllm.example.com/v1",
-            )
-        except Exception:
-            pass
-
-        mock_client.post.assert_called_once()
-        call_kwargs = mock_client.post.call_args[1]
-        request_data = json.loads(call_kwargs["data"])
-        assert request_data["messages"] == [
-            {
-                "role": "user",
-                "content": [
-                    {
-                        "type": "audio_url",
-                        "audio_url": {"url": "https://example.com/audio.mp3"},
-                    }
-                ],
-            }
-        ]
-
-
 def test_hosted_vllm_supports_reasoning_effort():
     config = HostedVLLMChatConfig()
-    supported_params = config.get_supported_openai_params(
-        model="hosted_vllm/gpt-oss-120b"
-    )
+    supported_params = config.get_supported_openai_params(model="hosted_vllm/gpt-oss-120b")
     assert "reasoning_effort" in supported_params
     optional_params = config.map_openai_params(
         non_default_params={"reasoning_effort": "high"},
@@ -129,9 +64,7 @@ def test_hosted_vllm_supports_thinking():
     Related issue: https://github.com/BerriAI/litellm/issues/19761
     """
     config = HostedVLLMChatConfig()
-    supported_params = config.get_supported_openai_params(
-        model="hosted_vllm/GLM-4.6-FP8"
-    )
+    supported_params = config.get_supported_openai_params(model="hosted_vllm/GLM-4.6-FP8")
     assert "thinking" in supported_params
 
     # Test thinking below the low threshold -> "minimal"
diff --git a/tests/test_litellm/llms/hosted_vllm/embedding/test_hosted_vllm_embedding_transformation.py b/tests/unit/llms/hosted_vllm/embedding/test_hosted_vllm_embedding_transformation.py
similarity index 97%
rename from tests/test_litellm/llms/hosted_vllm/embedding/test_hosted_vllm_embedding_transformation.py
rename to tests/unit/llms/hosted_vllm/embedding/test_hosted_vllm_embedding_transformation.py
index 34be3e12abd..5854b1596b4 100644
--- a/tests/test_litellm/llms/hosted_vllm/embedding/test_hosted_vllm_embedding_transformation.py
+++ b/tests/unit/llms/hosted_vllm/embedding/test_hosted_vllm_embedding_transformation.py
@@ -87,9 +87,7 @@ class TestHostedVLLMEmbeddingTransformation:
             headers={},
         )
 
-        assert (
-            "encoding_format" not in result
-        ), "encoding_format should not be in request when not provided"
+        assert "encoding_format" not in result, "encoding_format should not be in request when not provided"
 
     def test_encoding_format_not_included_when_none(self):
         """
@@ -278,9 +276,7 @@ class TestHostedVLLMEmbeddingTransformation:
             sent_data = json.loads(call_kwargs["data"])
 
             # Assert that encoding_format is NOT in the sent data
-            assert (
-                "encoding_format" not in sent_data
-            ), "encoding_format should not be in request when not provided"
+            assert "encoding_format" not in sent_data, "encoding_format should not be in request when not provided"
             assert sent_data["model"] == "BAAI/bge-small-en-v1.5"
             assert sent_data["input"] == ["Hello world"]
 
diff --git a/tests/test_litellm/llms/hosted_vllm/image_edit/test_hosted_vllm_image_edit_transformation.py b/tests/unit/llms/hosted_vllm/image_edit/test_hosted_vllm_image_edit_transformation.py
similarity index 100%
rename from tests/test_litellm/llms/hosted_vllm/image_edit/test_hosted_vllm_image_edit_transformation.py
rename to tests/unit/llms/hosted_vllm/image_edit/test_hosted_vllm_image_edit_transformation.py
diff --git a/tests/test_litellm/llms/hosted_vllm/responses/test_hosted_vllm_responses.py b/tests/unit/llms/hosted_vllm/responses/test_hosted_vllm_responses.py
similarity index 96%
rename from tests/test_litellm/llms/hosted_vllm/responses/test_hosted_vllm_responses.py
rename to tests/unit/llms/hosted_vllm/responses/test_hosted_vllm_responses.py
index e81bf0c4f1f..55d0ce1e68e 100644
--- a/tests/test_litellm/llms/hosted_vllm/responses/test_hosted_vllm_responses.py
+++ b/tests/unit/llms/hosted_vllm/responses/test_hosted_vllm_responses.py
@@ -68,9 +68,7 @@ def test_hosted_vllm_responses_create_with_string_input():
     Test that hosted_vllm routes directly to the native /v1/responses endpoint
     when the Responses API config is registered, and correctly parses the response.
     """
-    mock_client = _make_mock_http_client(
-        _make_mock_responses_api_response("I'm doing well, thanks!")
-    )
+    mock_client = _make_mock_http_client(_make_mock_responses_api_response("I'm doing well, thanks!"))
 
     with patch(
         "litellm.llms.custom_httpx.llm_http_handler._get_httpx_client",
@@ -109,10 +107,7 @@ def test_hosted_vllm_responses_create_with_explicit_none_extra_body():
     )
 
     # extra_body=None should be normalized to an empty dict (or absent)
-    assert (
-        optional_params.get("extra_body") is not None
-        or "extra_body" not in optional_params
-    )
+    assert optional_params.get("extra_body") is not None or "extra_body" not in optional_params
 
 
 def test_hosted_vllm_provider_config_registration():
diff --git a/tests/test_litellm/llms/hosted_vllm/test_hosted_vllm_rerank_transformation.py b/tests/unit/llms/hosted_vllm/test_hosted_vllm_rerank_transformation.py
similarity index 100%
rename from tests/test_litellm/llms/hosted_vllm/test_hosted_vllm_rerank_transformation.py
rename to tests/unit/llms/hosted_vllm/test_hosted_vllm_rerank_transformation.py
diff --git a/tests/test_litellm/llms/hosted_vllm/videos/test_hosted_vllm_video_transformation.py b/tests/unit/llms/hosted_vllm/videos/test_hosted_vllm_video_transformation.py
similarity index 100%
rename from tests/test_litellm/llms/hosted_vllm/videos/test_hosted_vllm_video_transformation.py
rename to tests/unit/llms/hosted_vllm/videos/test_hosted_vllm_video_transformation.py
diff --git a/tests/test_litellm/llms/huggingface/rerank/test_huggingface_rerank_transformation.py b/tests/unit/llms/huggingface/rerank/test_huggingface_rerank_transformation.py
similarity index 91%
rename from tests/test_litellm/llms/huggingface/rerank/test_huggingface_rerank_transformation.py
rename to tests/unit/llms/huggingface/rerank/test_huggingface_rerank_transformation.py
index 9d6b7290eb6..6fd2b006fef 100644
--- a/tests/test_litellm/llms/huggingface/rerank/test_huggingface_rerank_transformation.py
+++ b/tests/unit/llms/huggingface/rerank/test_huggingface_rerank_transformation.py
@@ -219,29 +219,6 @@ def test_huggingface_rerank_return_documents(mock_post):
             assert "text" in result["document"]
 
 
-@patch("litellm.llms.custom_httpx.http_handler.HTTPHandler.post")
-def test_huggingface_rerank_error_handling(mock_post):
-    """Test HuggingFace rerank error handling."""
-
-    def return_val():
-        return {"error": "Unauthorized"}
-
-    mock_response = MagicMock()
-    mock_response.status_code = 401
-    mock_response.json = return_val
-    mock_response.text = "Unauthorized"
-    mock_post.return_value = mock_response
-
-    with pytest.raises(litellm.APIConnectionError):
-        litellm.rerank(
-            model="huggingface/BAAI/bge-reranker-base",
-            query="hello",
-            documents=["hello", "world"],
-            top_n=2,
-            api_key="invalid_key",
-        )
-
-
 def test_huggingface_rerank_config():
     """Test HuggingFaceRerankConfig class functionality."""
     from litellm.llms.huggingface.rerank.transformation import HuggingFaceRerankConfig
@@ -249,10 +226,7 @@ def test_huggingface_rerank_config():
     config = HuggingFaceRerankConfig()
 
     # Test complete URL generation
-    assert (
-        config.get_complete_url(None, "test")
-        == "https://api-inference.huggingface.co/rerank"
-    )
+    assert config.get_complete_url(None, "test") == "https://api-inference.huggingface.co/rerank"
 
     # Test custom API base
     custom_url = config.get_complete_url("https://custom.huggingface.co", "test")
@@ -292,13 +266,9 @@ def test_request_transformation():
 
     config = HuggingFaceRerankConfig()
 
-    optional_params = OptionalRerankParams(
-        query="hello", texts=["hello", "world"], top_n=2, return_text=True
-    )
+    optional_params = OptionalRerankParams(query="hello", texts=["hello", "world"], top_n=2, return_text=True)
 
-    request_body = config.transform_rerank_request(
-        model="test", optional_rerank_params=optional_params, headers={}
-    )
+    request_body = config.transform_rerank_request(model="test", optional_rerank_params=optional_params, headers={})
 
     assert request_body["query"] == "hello"
     assert request_body["texts"] == ["hello", "world"]
@@ -368,9 +338,7 @@ def test_validate_environment():
 
     # Test headers override
     custom_headers = {"custom": "header"}
-    headers = config.validate_environment(
-        headers=custom_headers, model="test", api_key="test_key"
-    )
+    headers = config.validate_environment(headers=custom_headers, model="test", api_key="test_key")
 
     assert "custom" in headers
     assert headers["custom"] == "header"
diff --git a/tests/test_litellm/llms/inception/test_inception_chat_transformation.py b/tests/unit/llms/inception/test_inception_chat_transformation.py
similarity index 96%
rename from tests/test_litellm/llms/inception/test_inception_chat_transformation.py
rename to tests/unit/llms/inception/test_inception_chat_transformation.py
index 1d12be2adee..c4c023077fc 100644
--- a/tests/test_litellm/llms/inception/test_inception_chat_transformation.py
+++ b/tests/unit/llms/inception/test_inception_chat_transformation.py
@@ -188,21 +188,15 @@ def test_inception_does_not_leak_key_to_caller_api_base():
     caller also supplies their own key.
     """
     config = InceptionChatConfig()
-    with mock.patch.dict(
-        os.environ, {"INCEPTION_API_KEY": "server-secret"}, clear=True
-    ):
+    with mock.patch.dict(os.environ, {"INCEPTION_API_KEY": "server-secret"}, clear=True):
         with mock.patch.object(litellm, "inception_key", "module-secret"):
             # caller overrides api_base without a key -> server key withheld
-            api_base, api_key = config._get_openai_compatible_provider_info(
-                "https://attacker.example/v1", None
-            )
+            api_base, api_key = config._get_openai_compatible_provider_info("https://attacker.example/v1", None)
             assert api_base == "https://attacker.example/v1"
             assert api_key is None
 
             # caller overrides api_base AND supplies their own key -> used as-is
-            _, api_key = config._get_openai_compatible_provider_info(
-                "https://attacker.example/v1", "caller-key"
-            )
+            _, api_key = config._get_openai_compatible_provider_info("https://attacker.example/v1", "caller-key")
             assert api_key == "caller-key"
 
             # default/server base -> server-managed key resolved
@@ -217,9 +211,7 @@ def test_get_llm_provider_inception():
     assert model == "mercury-2"
     assert provider == "inception"
 
-    model, provider, _, api_base = get_llm_provider(
-        "mercury-2", api_base="https://api.inceptionlabs.ai/v1"
-    )
+    model, provider, _, api_base = get_llm_provider("mercury-2", api_base="https://api.inceptionlabs.ai/v1")
     assert model == "mercury-2"
     assert provider == "inception"
     assert api_base == "https://api.inceptionlabs.ai/v1"
@@ -293,5 +285,3 @@ def test_inception_completion_targets_inception_endpoint():
     assert captured["body"]["model"] == "mercury-2"
     assert captured["body"]["tool_choice"] == "auto"
     assert response.choices[0].message.content == "hi"
-
-
diff --git a/tests/test_litellm/llms/inception/test_inception_completion_transformation.py b/tests/unit/llms/inception/test_inception_completion_transformation.py
similarity index 95%
rename from tests/test_litellm/llms/inception/test_inception_completion_transformation.py
rename to tests/unit/llms/inception/test_inception_completion_transformation.py
index ed3f34fc744..84923229e20 100644
--- a/tests/test_litellm/llms/inception/test_inception_completion_transformation.py
+++ b/tests/unit/llms/inception/test_inception_completion_transformation.py
@@ -22,9 +22,7 @@ def _fim_response_bytes():
             "object": "text_completion",
             "created": 1,
             "model": "mercury-edit-2",
-            "choices": [
-                {"text": "a + b", "index": 0, "finish_reason": "stop", "logprobs": None}
-            ],
+            "choices": [{"text": "a + b", "index": 0, "finish_reason": "stop", "logprobs": None}],
             "usage": {"prompt_tokens": 5, "completion_tokens": 3, "total_tokens": 8},
         }
     ).encode()
@@ -47,9 +45,7 @@ def test_inception_fim_supports_suffix_param():
 
 def test_inception_fim_supported_params_match_schema():
     """FIM exposes the OpenAI subset of Inception's FIMCompletionRequest only"""
-    params = InceptionTextCompletionConfig().get_supported_openai_params(
-        "mercury-edit-2"
-    )
+    params = InceptionTextCompletionConfig().get_supported_openai_params("mercury-edit-2")
     for p in ("suffix", "top_p", "frequency_penalty", "presence_penalty", "stop"):
         assert p in params
     # Chat-only sampling controls are not part of Inception's FIM schema
@@ -75,11 +71,7 @@ def test_inception_get_supported_openai_params_dispatch():
 
 @pytest.mark.parametrize("provider", ["inception", "text-completion-inception"])
 def test_inception_validate_environment(provider):
-    model = (
-        "inception/mercury-2"
-        if provider == "inception"
-        else "text-completion-inception/mercury-edit-2"
-    )
+    model = "inception/mercury-2" if provider == "inception" else "text-completion-inception/mercury-edit-2"
 
     with mock.patch.dict(os.environ, {}, clear=True):
         result = litellm.validate_environment(model)
@@ -217,9 +209,7 @@ def test_inception_fim_does_not_leak_global_api_key():
             content=_fim_response_bytes(),
         )
 
-    with mock.patch.dict(
-        os.environ, {"INCEPTION_API_KEY": "sk-inception-correct"}, clear=True
-    ):
+    with mock.patch.dict(os.environ, {"INCEPTION_API_KEY": "sk-inception-correct"}, clear=True):
         with mock.patch.object(litellm, "inception_key", None):
             with mock.patch.object(litellm, "api_key", "sk-global-should-not-leak"):
                 with mock.patch("httpx.Client.send", new=fake_send):
diff --git a/tests/test_litellm/llms/jina_ai/embedding/test_jina_embedding_transformation.py b/tests/unit/llms/jina_ai/embedding/test_jina_embedding_transformation.py
similarity index 100%
rename from tests/test_litellm/llms/jina_ai/embedding/test_jina_embedding_transformation.py
rename to tests/unit/llms/jina_ai/embedding/test_jina_embedding_transformation.py
diff --git a/tests/test_litellm/llms/langflow/chat/test_langflow_chat_transformation.py b/tests/unit/llms/langflow/chat/test_langflow_chat_transformation.py
similarity index 93%
rename from tests/test_litellm/llms/langflow/chat/test_langflow_chat_transformation.py
rename to tests/unit/llms/langflow/chat/test_langflow_chat_transformation.py
index 383a7afbe93..179a6cad4aa 100644
--- a/tests/test_litellm/llms/langflow/chat/test_langflow_chat_transformation.py
+++ b/tests/unit/llms/langflow/chat/test_langflow_chat_transformation.py
@@ -46,7 +46,7 @@ def test_langflow_config_get_complete_url():
 
 def test_langflow_config_get_complete_url_requires_api_base():
     config = LangFlowConfig()
-    with pytest.raises(ValueError, match='api_base is required for LangFlow\\. Set it via'):
+    with pytest.raises(ValueError, match="api_base is required for LangFlow\\. Set it via"):
         config.get_complete_url(
             api_base=None,
             api_key=None,
@@ -225,9 +225,7 @@ def test_langflow_extra_body_cannot_inject_tweaks_into_run_payload():
         posted_bodies.append(json.loads(body) if isinstance(body, str) else body)
         resp = MagicMock(spec=httpx.Response)
         resp.status_code = 200
-        resp.json.return_value = {
-            "outputs": [{"outputs": [{"results": {"message": {"text": "hi"}}}]}]
-        }
+        resp.json.return_value = {"outputs": [{"outputs": [{"results": {"message": {"text": "hi"}}}]}]}
         resp.headers = {}
         resp.text = "{}"
         return resp
@@ -275,9 +273,7 @@ def test_langflow_config_extract_response_from_outputs_dict():
                     "outputs": [
                         {
                             "results": {},
-                            "outputs": {
-                                "message": {"message": {"text": "via outputs dict"}}
-                            },
+                            "outputs": {"message": {"message": {"text": "via outputs dict"}}},
                         }
                     ]
                 }
@@ -292,14 +288,9 @@ def test_langflow_extract_response_returns_none_when_no_message():
     assert config._extract_content_from_response({"outputs": []}) is None
     assert config._extract_content_from_response({"detail": "flow failed"}) is None
     assert config._extract_content_from_response({"outputs": ["not-a-dict"]}) is None
+    assert config._extract_content_from_response({"outputs": [{"outputs": ["bad"]}]}) is None
     assert (
-        config._extract_content_from_response({"outputs": [{"outputs": ["bad"]}]})
-        is None
-    )
-    assert (
-        config._extract_content_from_response(
-            {"outputs": [{"outputs": [{"results": {"message": {"text": ""}}}]}]}
-        )
+        config._extract_content_from_response({"outputs": [{"outputs": [{"results": {"message": {"text": ""}}}]}]})
         is None
     )
 
@@ -310,9 +301,7 @@ def test_langflow_transform_response_builds_model_response_with_usage():
         status_code=200,
         json={
             "session_id": "sess-abc",
-            "outputs": [
-                {"outputs": [{"results": {"message": {"text": "Hello from LangFlow"}}}]}
-            ],
+            "outputs": [{"outputs": [{"results": {"message": {"text": "Hello from LangFlow"}}}]}],
         },
     )
 
@@ -332,9 +321,7 @@ def test_langflow_transform_response_builds_model_response_with_usage():
     assert result.choices[0].finish_reason == "stop"
     assert result.model == "langflow/my-flow-id"
     assert result.usage.completion_tokens > 0
-    assert result.usage.total_tokens == (
-        result.usage.prompt_tokens + result.usage.completion_tokens
-    )
+    assert result.usage.total_tokens == (result.usage.prompt_tokens + result.usage.completion_tokens)
 
 
 def test_langflow_transform_response_raises_on_unparseable_body():
@@ -357,9 +344,7 @@ def test_langflow_transform_response_raises_on_unparseable_body():
 
 def test_langflow_transform_response_raises_on_non_json_body():
     config = LangFlowConfig()
-    raw_response = httpx.Response(
-        status_code=200, content=b"not json", headers={"content-type": "text/plain"}
-    )
+    raw_response = httpx.Response(status_code=200, content=b"not json", headers={"content-type": "text/plain"})
 
     with pytest.raises(LangFlowError):
         config.transform_response(
diff --git a/tests/test_litellm/llms/litellm_proxy/test_sandbox_executor.py b/tests/unit/llms/litellm_proxy/test_sandbox_executor.py
similarity index 84%
rename from tests/test_litellm/llms/litellm_proxy/test_sandbox_executor.py
rename to tests/unit/llms/litellm_proxy/test_sandbox_executor.py
index 422e7a3cf4d..e7a03b9231a 100644
--- a/tests/test_litellm/llms/litellm_proxy/test_sandbox_executor.py
+++ b/tests/unit/llms/litellm_proxy/test_sandbox_executor.py
@@ -55,9 +55,7 @@ def _install_fake_sandbox(monkeypatch, session_cls=_FakeSandboxSession):
 def test_execute_installs_inline_requirements_file(monkeypatch):
     _install_fake_sandbox(monkeypatch)
     executor = SkillsSandboxExecutor()
-    monkeypatch.setattr(
-        executor, "_collect_generated_files", lambda *args, **kwargs: []
-    )
+    monkeypatch.setattr(executor, "_collect_generated_files", lambda *args, **kwargs: [])
 
     requirements = "git+https://example.com/repo.git#egg=foo\n-r extra.txt\n-e ./pkg\n"
     result = executor.execute(
@@ -69,22 +67,15 @@ def test_execute_installs_inline_requirements_file(monkeypatch):
     assert result["success"] is True
 
     created_session = _FakeSandboxSession.last_instance
-    assert created_session.copied_contents[
-        "/sandbox/.litellm_requirements.txt"
-    ] == requirements.encode("utf-8")
-    assert (
-        "pip', 'install', '-r', '.litellm_requirements.txt'"
-        in created_session.run_calls[0]
-    )
+    assert created_session.copied_contents["/sandbox/.litellm_requirements.txt"] == requirements.encode("utf-8")
+    assert "pip', 'install', '-r', '.litellm_requirements.txt'" in created_session.run_calls[0]
     assert "os.chdir('/sandbox')" in created_session.run_calls[1]
 
 
 def test_execute_uses_skill_requirements_txt(monkeypatch):
     _install_fake_sandbox(monkeypatch)
     executor = SkillsSandboxExecutor()
-    monkeypatch.setattr(
-        executor, "_collect_generated_files", lambda *args, **kwargs: []
-    )
+    monkeypatch.setattr(executor, "_collect_generated_files", lambda *args, **kwargs: [])
 
     result = executor.execute(
         code="print('hello')",
@@ -97,9 +88,7 @@ def test_execute_uses_skill_requirements_txt(monkeypatch):
     assert result["success"] is True
 
     created_session = _FakeSandboxSession.last_instance
-    copied_paths = {
-        sandbox_path for _, sandbox_path in created_session.copy_to_runtime_calls
-    }
+    copied_paths = {sandbox_path for _, sandbox_path in created_session.copy_to_runtime_calls}
     assert "/sandbox/requirements.txt" in copied_paths
     assert "/sandbox/.litellm_requirements.txt" not in copied_paths
     assert "pip', 'install', '-r', 'requirements.txt'" in created_session.run_calls[0]
@@ -118,9 +107,7 @@ def test_execute_returns_install_failure(monkeypatch):
 
     _install_fake_sandbox(monkeypatch, session_cls=_FailingSandboxSession)
     executor = SkillsSandboxExecutor()
-    monkeypatch.setattr(
-        executor, "_collect_generated_files", lambda *args, **kwargs: []
-    )
+    monkeypatch.setattr(executor, "_collect_generated_files", lambda *args, **kwargs: [])
 
     result = executor.execute(
         code="print('hello')",
diff --git a/tests/test_litellm/llms/litellm_proxy/test_skills_ownership.py b/tests/unit/llms/litellm_proxy/test_skills_ownership.py
similarity index 88%
rename from tests/test_litellm/llms/litellm_proxy/test_skills_ownership.py
rename to tests/unit/llms/litellm_proxy/test_skills_ownership.py
index e538c50cde8..6caa2da3169 100644
--- a/tests/test_litellm/llms/litellm_proxy/test_skills_ownership.py
+++ b/tests/unit/llms/litellm_proxy/test_skills_ownership.py
@@ -37,12 +37,7 @@ def _skill(skill_id: str, created_by: str | None) -> LiteLLM_SkillsTable:
 def test_should_extract_skill_auth_from_supported_metadata_fields():
     auth = UserAPIKeyAuth(user_id="user-1")
 
-    assert (
-        skills_main._get_user_api_key_auth_from_kwargs(
-            {"metadata": {"user_api_key_auth": auth}}
-        )
-        is auth
-    )
+    assert skills_main._get_user_api_key_auth_from_kwargs({"metadata": {"user_api_key_auth": auth}}) is auth
     assert (
         skills_main._get_user_api_key_auth_from_kwargs(
             {"metadata": {}, "litellm_metadata": {"user_api_key_auth": auth}}
@@ -122,9 +117,7 @@ def test_should_forward_skill_auth_through_sdk_entrypoints(monkeypatch):
         == "deleted"
     )
 
-    assert handler.create_skill_handler.call_args.kwargs["metadata"] == {
-        "source": "request"
-    }
+    assert handler.create_skill_handler.call_args.kwargs["metadata"] == {"source": "request"}
     assert handler.create_skill_handler.call_args.kwargs["user_api_key_dict"] is auth
     assert handler.list_skills_handler.call_args.kwargs["user_api_key_dict"] is auth
     assert handler.get_skill_handler.call_args.kwargs["user_api_key_dict"] is auth
@@ -149,9 +142,7 @@ def test_should_build_resource_owner_scopes_for_auth_context():
     ]
     assert resource_ownership.get_primary_resource_owner_scope(auth) == "user-1"
     assert resource_ownership.user_can_access_resource_owner("team:team-1", auth)
-    assert resource_ownership.get_resource_owner_scopes(
-        UserAPIKeyAuth(token="token-hash")
-    ) == ["key:token-hash"]
+    assert resource_ownership.get_resource_owner_scopes(UserAPIKeyAuth(token="token-hash")) == ["key:token-hash"]
     # Identity-less callers get an empty scope set — sharing a sentinel
     # would collapse every identity-less caller into the same logical
     # owner, which is a cross-tenant data-access primitive.
@@ -165,9 +156,7 @@ def test_should_allow_admin_and_anonymous_resource_owner_paths():
     assert resource_ownership.is_proxy_admin(admin)
     assert resource_ownership.user_can_access_resource_owner(None, admin)
     assert resource_ownership.user_can_access_resource_owner(None, None)
-    assert not resource_ownership.user_can_access_resource_owner(
-        None, UserAPIKeyAuth(user_id="user-1")
-    )
+    assert not resource_ownership.user_can_access_resource_owner(None, UserAPIKeyAuth(user_id="user-1"))
 
 
 @pytest.mark.asyncio
@@ -218,9 +207,7 @@ async def test_should_forward_skill_auth_through_transformation_handler(monkeypa
 async def test_should_store_team_owner_for_keys_without_user_id(monkeypatch):
     table = AsyncMock()
     table.create.side_effect = lambda data: _skill(data["skill_id"], data["created_by"])
-    prisma_client = type(
-        "Prisma", (), {"db": type("DB", (), {"litellm_skillstable": table})()}
-    )()
+    prisma_client = type("Prisma", (), {"db": type("DB", (), {"litellm_skillstable": table})()})()
     monkeypatch.setattr(
         LiteLLMSkillsHandler,
         "_get_prisma_client",
@@ -242,9 +229,7 @@ async def test_should_store_team_owner_for_keys_without_user_id(monkeypatch):
 async def test_should_store_token_owner_for_keys_without_user_team_or_org(monkeypatch):
     table = AsyncMock()
     table.create.side_effect = lambda data: _skill(data["skill_id"], data["created_by"])
-    prisma_client = type(
-        "Prisma", (), {"db": type("DB", (), {"litellm_skillstable": table})()}
-    )()
+    prisma_client = type("Prisma", (), {"db": type("DB", (), {"litellm_skillstable": table})()})()
     monkeypatch.setattr(
         LiteLLMSkillsHandler,
         "_get_prisma_client",
@@ -268,9 +253,7 @@ async def test_should_reject_skill_create_for_identityless_proxy_auth(monkeypatc
     sentinel as ``created_by`` would let any two such callers see each
     other's skills via the resulting shared owner scope."""
     table = AsyncMock()
-    prisma_client = type(
-        "Prisma", (), {"db": type("DB", (), {"litellm_skillstable": table})()}
-    )()
+    prisma_client = type("Prisma", (), {"db": type("DB", (), {"litellm_skillstable": table})()})()
     monkeypatch.setattr(
         LiteLLMSkillsHandler,
         "_get_prisma_client",
@@ -291,9 +274,7 @@ async def test_should_reject_skill_create_for_identityless_proxy_auth(monkeypatc
 async def test_should_filter_list_skills_to_authenticated_owner_scopes(monkeypatch):
     table = AsyncMock()
     table.find_many.return_value = [_skill("litellm_skill_owner", "user-1")]
-    prisma_client = type(
-        "Prisma", (), {"db": type("DB", (), {"litellm_skillstable": table})()}
-    )()
+    prisma_client = type("Prisma", (), {"db": type("DB", (), {"litellm_skillstable": table})()})()
     monkeypatch.setattr(
         LiteLLMSkillsHandler,
         "_get_prisma_client",
@@ -318,9 +299,7 @@ async def test_should_filter_list_skills_to_authenticated_owner_scopes(monkeypat
 async def test_should_hide_skill_from_different_owner(monkeypatch):
     table = AsyncMock()
     table.find_unique.return_value = _skill("litellm_skill_other", "user-2")
-    prisma_client = type(
-        "Prisma", (), {"db": type("DB", (), {"litellm_skillstable": table})()}
-    )()
+    prisma_client = type("Prisma", (), {"db": type("DB", (), {"litellm_skillstable": table})()})()
     monkeypatch.setattr(
         LiteLLMSkillsHandler,
         "_get_prisma_client",
@@ -340,9 +319,7 @@ async def test_should_hide_skill_from_different_owner(monkeypatch):
 async def test_should_hide_unowned_skill_by_default(monkeypatch):
     table = AsyncMock()
     table.find_unique.return_value = _skill("litellm_skill_unowned", None)
-    prisma_client = type(
-        "Prisma", (), {"db": type("DB", (), {"litellm_skillstable": table})()}
-    )()
+    prisma_client = type("Prisma", (), {"db": type("DB", (), {"litellm_skillstable": table})()})()
     monkeypatch.setattr(
         LiteLLMSkillsHandler,
         "_get_prisma_client",
@@ -364,9 +341,7 @@ async def test_list_skills_excludes_unowned_for_non_admin(monkeypatch):
     with ``created_by IS NULL`` are excluded — admin-only."""
     table = AsyncMock()
     table.find_many.return_value = []
-    prisma_client = type(
-        "Prisma", (), {"db": type("DB", (), {"litellm_skillstable": table})()}
-    )()
+    prisma_client = type("Prisma", (), {"db": type("DB", (), {"litellm_skillstable": table})()})()
     monkeypatch.setattr(
         LiteLLMSkillsHandler,
         "_get_prisma_client",
@@ -422,9 +397,7 @@ async def test_load_skill_uses_cache_after_first_db_hit(monkeypatch):
     fake_skill = Mock(created_by="user-1", skill_id="litellm_skill_a")
     table = AsyncMock()
     table.find_unique = AsyncMock(return_value=fake_skill)
-    prisma_client = type(
-        "Prisma", (), {"db": type("DB", (), {"litellm_skillstable": table})()}
-    )()
+    prisma_client = type("Prisma", (), {"db": type("DB", (), {"litellm_skillstable": table})()})()
     monkeypatch.setattr(
         skills_handler.LiteLLMSkillsHandler,
         "_get_prisma_client",
@@ -432,10 +405,7 @@ async def test_load_skill_uses_cache_after_first_db_hit(monkeypatch):
     )
 
     for _ in range(3):
-        assert (
-            await skills_handler.LiteLLMSkillsHandler._load_skill("litellm_skill_a")
-            is fake_skill
-        )
+        assert await skills_handler.LiteLLMSkillsHandler._load_skill("litellm_skill_a") is fake_skill
     assert table.find_unique.await_count == 1
 
 
@@ -445,9 +415,7 @@ async def test_load_skill_caches_negative_lookups(monkeypatch):
     the DB and the caller still sees ``None``."""
     table = AsyncMock()
     table.find_unique = AsyncMock(return_value=None)
-    prisma_client = type(
-        "Prisma", (), {"db": type("DB", (), {"litellm_skillstable": table})()}
-    )()
+    prisma_client = type("Prisma", (), {"db": type("DB", (), {"litellm_skillstable": table})()})()
     monkeypatch.setattr(
         skills_handler.LiteLLMSkillsHandler,
         "_get_prisma_client",
@@ -466,9 +434,7 @@ async def test_delete_skill_invalidates_cache(monkeypatch):
     table = AsyncMock()
     table.find_unique = AsyncMock(return_value=fake_skill)
     table.delete = AsyncMock()
-    prisma_client = type(
-        "Prisma", (), {"db": type("DB", (), {"litellm_skillstable": table})()}
-    )()
+    prisma_client = type("Prisma", (), {"db": type("DB", (), {"litellm_skillstable": table})()})()
     monkeypatch.setattr(
         skills_handler.LiteLLMSkillsHandler,
         "_get_prisma_client",
@@ -480,12 +446,7 @@ async def test_delete_skill_invalidates_cache(monkeypatch):
     assert skills_handler._SKILL_CACHE.get_cache("litellm_skill_a") is fake_skill
 
     auth = UserAPIKeyAuth(user_id="user-1")
-    await skills_handler.LiteLLMSkillsHandler.delete_skill(
-        "litellm_skill_a", user_api_key_dict=auth
-    )
+    await skills_handler.LiteLLMSkillsHandler.delete_skill("litellm_skill_a", user_api_key_dict=auth)
 
     # Post-delete, the cache holds the negative sentinel — not the stale row.
-    assert (
-        skills_handler._SKILL_CACHE.get_cache("litellm_skill_a")
-        == skills_handler._NEGATIVE_SKILL_SENTINEL
-    )
+    assert skills_handler._SKILL_CACHE.get_cache("litellm_skill_a") == skills_handler._NEGATIVE_SKILL_SENTINEL