test(lint): ban blind pytest.raises(Exception) with ruff B017 (#37731)

* test(lint): ban blind pytest.raises(Exception) with ruff B017

A bare pytest.raises(Exception) accepts whatever the body throws. The TypeError
a refactor introduces satisfies it exactly as well as the rejection the test was
written for, so the crash reads as a pass and the test never goes red.

All 111 existing sites are narrowed here. A runtime probe recorded the concrete
exception each one actually catches, and each site now names that type. Where
the code under test genuinely raises a bare Exception, the site pins a stable
slice of the message with match= instead.

Two sites tell on themselves. The shared responses-API cancel test raises
"custom_llm_provider is required but passed as None" rather than talking to a
provider at all, because cancel_responses takes a provider, not a model. And
test_bedrock_guardrails_with_streaming was the only test in its file still
passing without AWS credentials, because the NoCredentialsError boto3 raised
long before the guardrail ran satisfied the blind raises.

* fix(test): widen the openai batch-dispatch assertion to OpenAIError

The narrowed NotFoundError only holds where OPENAI_API_KEY is set. Without one
the SDK raises OpenAIError while building the client, long before any 404, so CI
went red. OpenAIError covers both and still rejects a TypeError from a refactor.

ruff-tests.toml

@@ -14,6 +14,9 @@
 # B018    a bare attribute access or literal, usually a call missing its parens
 # PLW0127 `x = x` self-assignment, dead code that reads like a narrowing or a fixup
 # PLR0133 comparison of two constants, e.g. `assert True == True`
+# B017    `pytest.raises(Exception)` accepts the TypeError a refactor introduced just as
+#         readily as the rejection under test, so a crash reads as a pass. Narrow to the
+#         real type, or add `match=` where the code genuinely raises a bare Exception
 #
 # No target-version here on purpose: it resolves from requires-python (>=3.10), so
 # 3.11-only builtins like BaseExceptionGroup are correctly flagged in a tree that
@@ -21,4 +24,4 @@
 
 line-length = 120
 
-lint.select = ["F821", "B011", "B015", "B018", "PT015", "PLR0133", "PLW0127"]
+lint.select = ["F821", "B011", "B015", "B017", "B018", "PT015", "PLR0133", "PLW0127"]

tests/guardrails_tests/test_bedrock_guardrails.py

@@ -197,6 +197,7 @@ async def test_bedrock_guardrails_block_responses_api():
 
 @pytest.mark.asyncio
 async def test_bedrock_guardrails_with_streaming():
+    from fastapi import HTTPException
     from litellm.proxy.utils import ProxyLogging
     from litellm.types.guardrails import GuardrailEventHooks
 
@@ -204,7 +205,7 @@ async def test_bedrock_guardrails_with_streaming():
     mock_user_api_key_cache = MagicMock(spec=DualCache)
     mock_user_api_key_dict = UserAPIKeyAuth()
 
-    with pytest.raises(Exception):  # Assert that this raises an exception
+    with pytest.raises(HTTPException):
         proxy_logging_obj = ProxyLogging(
             user_api_key_cache=mock_user_api_key_cache,
             premium_user=True,

tests/guardrails_tests/test_eu_ai_act_article5.py

@@ -20,6 +20,7 @@ from litellm.proxy.guardrails.guardrail_hooks.litellm_content_filter.content_fil
 from litellm.types.proxy.guardrails.guardrail_hooks.litellm_content_filter import (
     ContentFilterCategoryConfig,
 )
+from fastapi import HTTPException
 
 
 # Test cases: (sentence, expected_result, reason)
@@ -275,7 +276,7 @@ class TestEUAIActEdgeCases:
         for sentence in sentences:
             request_data = {"messages": [{"role": "user", "content": sentence}]}
 
-            with pytest.raises(Exception):
+            with pytest.raises(HTTPException):
                 await content_filter_guardrail.apply_guardrail(
                     inputs={"texts": [sentence]},
                     request_data=request_data,
@@ -289,7 +290,7 @@ class TestEUAIActEdgeCases:
         request_data = {"messages": [{"role": "user", "content": sentence}]}
 
         # Should block (contains multiple violations)
-        with pytest.raises(Exception):
+        with pytest.raises(HTTPException):
             await content_filter_guardrail.apply_guardrail(
                 inputs={"texts": [sentence]},
                 request_data=request_data,

tests/guardrails_tests/test_eu_ai_act_french_3_scenarios.py

@@ -19,6 +19,7 @@ from litellm.proxy.guardrails.guardrail_hooks.litellm_content_filter.content_fil
 from litellm.types.proxy.guardrails.guardrail_hooks.litellm_content_filter import (
     ContentFilterCategoryConfig,
 )
+from fastapi import HTTPException
 
 
 @pytest.fixture
@@ -228,7 +229,7 @@ class TestFrenchEdgeCases:
         request_data = {"messages": [{"role": "user", "content": sentence}]}
 
         # Should block (contains "build" and "système de crédit social")
-        with pytest.raises(Exception):
+        with pytest.raises(HTTPException):
             await content_filter_guardrail.apply_guardrail(
                 inputs={"texts": [sentence]},
                 request_data=request_data,
@@ -257,7 +258,7 @@ class TestFrenchEdgeCases:
         request_data = {"messages": [{"role": "user", "content": sentence}]}
 
         # Should block (case-insensitive)
-        with pytest.raises(Exception):
+        with pytest.raises(HTTPException):
             await content_filter_guardrail.apply_guardrail(
                 inputs={"texts": [sentence]},
                 request_data=request_data,

tests/guardrails_tests/test_semantic_guard.py

@@ -10,6 +10,7 @@ sys.path.insert(0, os.path.abspath("../.."))
 from unittest.mock import MagicMock
 
 import pytest
+from fastapi import HTTPException
 
 
 class TestRouteLoader:
@@ -307,7 +308,7 @@ class TestContentFilterSqlInjectionTemplate:
     @pytest.mark.asyncio
     async def test_sql_always_block(self, sql_injection_guardrail, sentence, reason):
         request_data = {"messages": [{"role": "user", "content": sentence}]}
-        with pytest.raises(Exception):
+        with pytest.raises(HTTPException):
             await sql_injection_guardrail.apply_guardrail(
                 inputs={"texts": [sentence]},
                 request_data=request_data,
@@ -343,7 +344,7 @@ class TestContentFilterSqlInjectionTemplate:
         self, sql_injection_guardrail, sentence, reason
     ):
         request_data = {"messages": [{"role": "user", "content": sentence}]}
-        with pytest.raises(Exception):
+        with pytest.raises(HTTPException):
             await sql_injection_guardrail.apply_guardrail(
                 inputs={"texts": [sentence]},
                 request_data=request_data,
@@ -552,7 +553,7 @@ class TestContentFilterPromptInjectionTemplate:
     @pytest.mark.asyncio
     async def test_always_block(self, content_filter_guardrail, sentence, reason):
         request_data = {"messages": [{"role": "user", "content": sentence}]}
-        with pytest.raises(Exception):
+        with pytest.raises(HTTPException):
             await content_filter_guardrail.apply_guardrail(
                 inputs={"texts": [sentence]},
                 request_data=request_data,

tests/image_gen_tests/test_image_generation.py

@@ -444,7 +444,7 @@ async def test_azure_image_generation_request_body():
     ) as mock_post:
         mock_post.side_effect = Exception("test")
 
-        with pytest.raises(Exception):
+        with pytest.raises(litellm.APIConnectionError):
             await aimage_generation(
                 model="azure/gpt-image-1",
                 prompt="test prompt",

tests/litellm_utils_tests/test_utils.py

@@ -1334,7 +1334,7 @@ def test_validate_chat_completion_user_messages(messages, expected_bool):
         validate_chat_completion_user_messages(messages=messages)
     else:
         ## Invalid message
-        with pytest.raises(Exception):
+        with pytest.raises(Exception, match="Invalid user message at index 0"):
             validate_chat_completion_user_messages(messages=messages)
 
 
@@ -1354,7 +1354,7 @@ def test_validate_chat_completion_tool_choice(tool_choice, expected_bool):
     if expected_bool:
         validate_chat_completion_tool_choice(tool_choice=tool_choice)
     else:
-        with pytest.raises(Exception):
+        with pytest.raises(Exception, match="Invalid tool choice"):
             validate_chat_completion_tool_choice(tool_choice=tool_choice)
 
 

tests/llm_responses_api_testing/base_responses_api.py

@@ -28,6 +28,7 @@ from openai.types.responses.response_create_params import (
     ResponseInputParam,
 )
 from litellm.llms.custom_httpx.http_handler import AsyncHTTPHandler
+import openai
 
 
 def validate_responses_api_response(response, final_chunk: bool = False):
@@ -700,12 +701,12 @@ class BaseResponsesAPITest(ABC):
         base_completion_call_args = self.get_base_completion_call_args()
 
         if sync_mode:
-            with pytest.raises(Exception):
+            with pytest.raises(openai.APIError):
                 litellm.cancel_responses(
                     response_id="invalid_response_id_12345", **base_completion_call_args
                 )
         else:
-            with pytest.raises(Exception):
+            with pytest.raises(openai.APIError):
                 await litellm.acancel_responses(
                     response_id="invalid_response_id_12345", **base_completion_call_args
                 )

tests/llm_translation/test_llm_response_utils/test_convert_dict_to_chat_completion.py

@@ -1243,7 +1243,7 @@ def test_convert_to_model_response_object_with_error_code_only():
         },
     }
 
-    with pytest.raises(Exception):
+    with pytest.raises(Exception) as exc_info:  # noqa: B017  # bare Exception raised, so status_code is the assertion
         convert_to_model_response_object(
             model_response_object=ModelResponse(),
             response_object=response_object,
@@ -1255,6 +1255,8 @@ def test_convert_to_model_response_object_with_error_code_only():
             convert_tool_call_to_json_mode=False,
         )
 
+    assert exc_info.value.status_code == 500
+
 
 def test_model_prefix_preservation():
     """
@@ -2473,14 +2475,14 @@ class TestConvertToModelResponseObjectCompletion:
         assert "reasoning_content" not in (message.provider_specific_fields or {})
 
     def test_response_none_raises(self):
-        with pytest.raises(Exception):
+        with pytest.raises(Exception, match="Invalid response object"):
             convert_to_model_response_object(
                 response_object=None,
                 model_response_object=ModelResponse(),
             )
 
     def test_model_response_none_raises(self):
-        with pytest.raises(Exception):
+        with pytest.raises(Exception, match="Invalid response object"):
             convert_to_model_response_object(
                 response_object={
                     "choices": [

tests/llm_translation/test_openai.py

@@ -1458,7 +1458,7 @@ def test_responses_gpt54_with_xhigh_reasoning():
         # Stop execution right after request generation to avoid external API calls.
         mock_responses.side_effect = RuntimeError("stop_after_request_build")
 
-        with pytest.raises(Exception):
+        with pytest.raises(litellm.APIConnectionError):
             litellm.completion(
                 model="openai/responses/gpt-5.4",
                 messages=[{"role": "user", "content": "What is 2+2?"}],

tests/local_testing/test_get_llm_provider.py

@@ -569,5 +569,5 @@ class TestClaudeModelPatternMatching:
         )
 
         set_fallback_generalizations([])
-        with pytest.raises(Exception):
+        with pytest.raises(litellm.BadRequestError):
             litellm.get_llm_provider(model="claude-opus-4-9")

tests/openai_endpoints_tests/test_e2e_openai_responses_api.py

@@ -1,5 +1,5 @@
 import httpx
-from openai import OpenAI, BadRequestError
+from openai import OpenAI, BadRequestError, APIStatusError
 import pytest
 
 
@@ -87,7 +87,7 @@ def test_basic_response():
     print("DELETE response=", delete_response)
 
     # expect an error when getting the response again since it was deleted
-    with pytest.raises(Exception):
+    with pytest.raises(APIStatusError):
         get_response = client.responses.retrieve(response.id)
 
 
@@ -195,6 +195,6 @@ def test_cancel_streaming_response():
 
 def test_cancel_invalid_response_id():
     client = get_test_client()
-    with pytest.raises(Exception):
+    with pytest.raises(APIStatusError):
         # Try to cancel a non-existent response ID
         client.responses.cancel("invalid_response_id_12345")

tests/proxy_unit_tests/test_auth_checks.py

@@ -330,6 +330,7 @@ async def test_wildcard_access_after_cost_map_reload(key_models, model, expect_t
     Fix: each reload now calls litellm.add_known_models(model_cost_map=new_map)
     with the fetched map passed explicitly to avoid any reference ambiguity.
     """
+    from litellm.proxy._types import ProxyException
     from litellm.proxy.auth.auth_checks import can_key_call_model
 
     # Build a new cost map that includes the brand-new model — exactly what
@@ -378,7 +379,7 @@ async def test_wildcard_access_after_cost_map_reload(key_models, model, expect_t
                 llm_router=router,
             )
         else:
-            with pytest.raises(Exception):
+            with pytest.raises(ProxyException):
                 await can_key_call_model(
                     model=model,
                     llm_model_list=llm_model_list,

tests/proxy_unit_tests/test_jwt.py

@@ -41,6 +41,7 @@ from litellm.proxy.auth.handle_jwt import JWTHandler, JWTAuthManager
 from litellm.proxy.management_endpoints.team_endpoints import new_team
 from litellm.proxy.proxy_server import chat_completion
 from typing import Literal, Optional
+from litellm.proxy._types import ProxyException
 
 public_key = {
     "kty": "RSA",
@@ -1045,9 +1046,7 @@ async def test_allow_access_by_email(
             assert result is not None  # Adjust this based on your actual response check
         else:
             # Expect the call to fail
-            with pytest.raises(
-                Exception
-            ):  # Replace with the actual exception raised on failure
+            with pytest.raises(ProxyException):
                 resp = await user_api_key_auth(request=request, api_key=bearer_token)
                 print(resp)
 

tests/proxy_unit_tests/test_proxy_config_unit_test.py

@@ -53,7 +53,7 @@ async def test_read_config_from_bad_file_path():
     """
     proxy_config_instance = ProxyConfig()
     config_path = "non-existent-file.yaml"
-    with pytest.raises(Exception):
+    with pytest.raises(Exception, match="Config file not found"):
         config = await proxy_config_instance.get_config(config_file_path=config_path)
 
 

tests/proxy_unit_tests/test_proxy_utils.py

@@ -29,6 +29,7 @@ from litellm.proxy.litellm_pre_call_utils import (
     _get_dynamic_logging_metadata,
     add_litellm_data_to_request,
 )
+from pydantic import ValidationError
 
 pytestmark = pytest.mark.xdist_group("proxy_heavy")
 
@@ -1695,13 +1696,13 @@ def test_update_key_request_validation():
     """
     from litellm.proxy._types import UpdateKeyRequest
 
-    with pytest.raises(Exception):
+    with pytest.raises(ValidationError):
         UpdateKeyRequest(
             key="test_key",
             temp_budget_increase=100,
         )
 
-    with pytest.raises(Exception):
+    with pytest.raises(ValidationError):
         UpdateKeyRequest(
             key="test_key",
             temp_budget_expiry="2024-01-20T00:00:00Z",
@@ -1848,7 +1849,7 @@ async def test_end_user_transactions_reset():
     mock_client.db.tx = AsyncMock(side_effect=Exception("DB Error"))
 
     # Call function - should raise error
-    with pytest.raises(Exception):
+    with pytest.raises(TypeError):
         await ProxyUpdateSpend.update_end_user_spend(
             n_retry_times=0,
             prisma_client=mock_client,
@@ -1878,7 +1879,7 @@ async def test_spend_logs_cleanup_after_error():
     original_logs = mock_client.spend_log_transactions.copy()
 
     # Call function - should raise error
-    with pytest.raises(Exception):
+    with pytest.raises(TypeError):
         await ProxyUpdateSpend.update_spend_logs(
             n_retry_times=0,
             prisma_client=mock_client,

tests/proxy_unit_tests/test_skills_db.py

@@ -26,6 +26,7 @@ from litellm.proxy import proxy_server
 from litellm.proxy._types import UserAPIKeyAuth
 from litellm.proxy.utils import PrismaClient, ProxyLogging
 from litellm.types.utils import LlmProviders
+import openai
 
 proxy_logging_obj = ProxyLogging(user_api_key_cache=DualCache())
 
@@ -254,7 +255,7 @@ async def test_delete_skill_sdk(prisma_client):
     assert result.type == "skill_deleted"
 
     # Verify skill no longer exists
-    with pytest.raises(Exception):
+    with pytest.raises(openai.APIError):
         await aget_skill(
             skill_id=created_skill.id,
             custom_llm_provider=LlmProviders.LITELLM_PROXY.value,

tests/proxy_unit_tests/test_user_api_key_auth.py

@@ -436,7 +436,7 @@ def test_ui_token_route_access(route, user_role, should_be_allowed):
         )
         assert result is True
     else:
-        with pytest.raises(Exception):
+        with pytest.raises(Exception, match="Only proxy admin can be used to generate"):
             _is_api_route_allowed(
                 route=route,
                 request=request,

tests/test_litellm/caching/test_redis_cache.py

@@ -520,13 +520,15 @@ async def test_circuit_breaker_covers_lua_script_execution(redis_no_ping):
     counted toward taking Redis out of the pool and kept paying a full socket timeout
     each, which is the traffic the outage hurts most.
     """
+    from redis.exceptions import ConnectionError as RedisConnectionError
+
     from litellm.constants import REDIS_CIRCUIT_BREAKER_FAILURE_THRESHOLD
 
     cache = RedisCache(host="127.0.0.1", port=_closed_port(), socket_timeout=0.5)
     run_script = cache.async_register_script("return 1")
 
     for _ in range(REDIS_CIRCUIT_BREAKER_FAILURE_THRESHOLD):
-        with pytest.raises(Exception):
+        with pytest.raises(RedisConnectionError):
             await run_script(keys=["lit4930"], args=[1])
 
     with pytest.raises(Exception, match="circuit breaker is open"):
@@ -603,8 +605,11 @@ async def test_only_connectivity_failures_open_the_breaker(error, opens_breaker)
     async def failing_call():
         raise raised
 
-    for _ in range(breaker.failure_threshold + 1):
-        with pytest.raises(Exception):
+    for _ in range(breaker.failure_threshold):
+        with pytest.raises(type(raised)):
             await _run_under_circuit_breaker(breaker, "op", failing_call)
 
+    with pytest.raises(Exception, match="circuit breaker is open" if opens_breaker else "boom"):
+        await _run_under_circuit_breaker(breaker, "op", failing_call)
+
     assert breaker.is_open() is opens_breaker

tests/test_litellm/containers/test_container_api.py

@@ -384,7 +384,7 @@ class TestContainerAPI:
             "container_create_handler",
             side_effect=Exception("API Error"),
         ):
-            with pytest.raises(Exception):
+            with pytest.raises(litellm.APIConnectionError):
                 create_container(
                     name="Error Test Container", custom_llm_provider="openai"
                 )

tests/test_litellm/interactions/test_agents_main_and_utils.py

@@ -314,7 +314,7 @@ class TestAsyncErrorWrapping:
         handler.create_agent.side_effect = RuntimeError("kaboom")
 
         with patch(_HANDLER_PATH, handler):
-            with pytest.raises(Exception):
+            with pytest.raises(litellm.APIConnectionError):
                 await acreate(name="waverunner", api_key="AIza")
 
     @pytest.mark.asyncio
@@ -323,7 +323,7 @@ class TestAsyncErrorWrapping:
         handler.get_agent.side_effect = RuntimeError("kaboom")
 
         with patch(_HANDLER_PATH, handler):
-            with pytest.raises(Exception):
+            with pytest.raises(litellm.APIConnectionError):
                 await aget(name="waverunner", api_key="AIza")
 
     @pytest.mark.asyncio
@@ -332,7 +332,7 @@ class TestAsyncErrorWrapping:
         handler.list_agents.side_effect = RuntimeError("kaboom")
 
         with patch(_HANDLER_PATH, handler):
-            with pytest.raises(Exception):
+            with pytest.raises(litellm.APIConnectionError):
                 await alist(api_key="AIza")
 
     @pytest.mark.asyncio
@@ -341,7 +341,7 @@ class TestAsyncErrorWrapping:
         handler.delete_agent.side_effect = RuntimeError("kaboom")
 
         with patch(_HANDLER_PATH, handler):
-            with pytest.raises(Exception):
+            with pytest.raises(litellm.APIConnectionError):
                 await adelete(name="waverunner", api_key="AIza")
 
     @pytest.mark.asyncio
@@ -350,5 +350,5 @@ class TestAsyncErrorWrapping:
         handler.list_agent_versions.side_effect = RuntimeError("kaboom")
 
         with patch(_HANDLER_PATH, handler):
-            with pytest.raises(Exception):
+            with pytest.raises(litellm.APIConnectionError):
                 await alist_versions(name="waverunner", api_key="AIza")

tests/test_litellm/interactions/test_google_interactions_integration.py

@@ -18,6 +18,7 @@ sys.path.insert(0, os.path.abspath("../../.."))
 
 import litellm
 import litellm.interactions as interactions
+import openai
 
 # Test API key - should be set in environment
 GEMINI_API_KEY = os.getenv("GEMINI_API_KEY")
@@ -258,7 +259,7 @@ class TestGoogleInteractionsErrorHandling:
 
     def test_invalid_model(self, api_key):
         """Test error handling for invalid model."""
-        with pytest.raises(Exception):
+        with pytest.raises(openai.APIError):
             interactions.create(
                 model="gemini/invalid-model-name-xyz",
                 input="Hello",
@@ -267,7 +268,7 @@ class TestGoogleInteractionsErrorHandling:
 
     def test_missing_model_and_agent(self, api_key):
         """Test error when neither model nor agent is provided."""
-        with pytest.raises(Exception):  # Can be ValueError or APIConnectionError
+        with pytest.raises((ValueError, litellm.APIConnectionError)):
             interactions.create(
                 input="Hello",
                 api_key=api_key,

tests/test_litellm/litellm_core_utils/test_fallback_generalizations.py

@@ -288,7 +288,7 @@ def test_capability_info_backfills_requested_provider(restore_generalizations):
 def test_routing_only_match_does_not_resolve_model_info(restore_generalizations):
     restore_generalizations([{"name": "route", "pattern": r"^ceeco-", "model_info": {"litellm_provider": "openai"}}])
     litellm.get_model_info.cache_clear()
-    with pytest.raises(Exception):
+    with pytest.raises(Exception, match="This model isn't mapped yet"):
         litellm.get_model_info("ceeco-fast-1", custom_llm_provider="openai")
 
 
@@ -470,7 +470,7 @@ def test_shipped_adaptive_rule_requires_claude_prefix(shipped_cost_map):
     model = "openai/team-sonnet-5-1-alias"
     assert model not in litellm.model_cost
     assert match_capability_generalizations("team-sonnet-5-1-alias") is None
-    with pytest.raises(Exception):
+    with pytest.raises(Exception, match="This model isn't mapped yet"):
         litellm.get_model_info(model)
 
 
@@ -496,7 +496,7 @@ def test_shipped_rules_lose_to_exact_entries_across_cost_ladder_variants(shipped
     from litellm.types.utils import ModelResponse, Usage
 
     assert "claude-haiku-4-5-20251001" in litellm.model_cost
-    with pytest.raises(Exception):
+    with pytest.raises(Exception, match="This model isn't mapped yet"):
         litellm.get_model_info("claude-haiku-4-5-20251001", custom_llm_provider="bedrock")
 
     entry = litellm.model_cost["us.anthropic.claude-haiku-4-5-20251001-v1:0"]

tests/test_litellm/litellm_core_utils/test_realtime_streaming.py

@@ -62,7 +62,7 @@ def test_realtime_streaming_store_message():
 
     # Test 3: Invalid message format
     invalid_msg = "invalid json"
-    with pytest.raises(Exception):
+    with pytest.raises(json.JSONDecodeError):
         streaming.store_message(invalid_msg)
 
     # Test 4: Message type not in logged events

tests/test_litellm/litellm_core_utils/test_streaming_handler.py

@@ -4176,7 +4176,7 @@ async def test_stream_wrapper_anext_max_duration_timeout_restores_consumer_corre
 
         wrapper._stream_created_time = time.time() - 10
 
-        with pytest.raises(Exception):
+        with pytest.raises(litellm.Timeout):
             await wrapper.__anext__()
 
         assert trace_id_var.get() == "outer-trace-max-duration"
@@ -4323,7 +4323,9 @@ def test_handle_stream_fallback_error_restores_context_only_after_exception_mapp
 
         monkeypatch.setattr("litellm.litellm_core_utils.streaming_handler.exception_type", fake_exception_type)
 
-        with pytest.raises(Exception):
+        from litellm.exceptions import MidStreamFallbackError
+
+        with pytest.raises(MidStreamFallbackError):
             wrapper._handle_stream_fallback_error(RuntimeError("boom"))
 
         # The mapper ran while the stream's own ids were still active.

tests/test_litellm/llms/anthropic/experimental_pass_through/messages/test_mcp_handler.py

@@ -61,7 +61,7 @@ def test_anthropic_messages_handler_skips_the_gateway_on_recursion():
         "litellm.llms.anthropic.experimental_pass_through.messages.mcp_handler.anthropic_messages_with_mcp",
         new=AsyncMock(return_value={"routed": True}),
     ) as routed:
-        with pytest.raises(Exception):
+        with pytest.raises(ValueError):
             anthropic_messages_handler(
                 max_tokens=100,
                 messages=[{"role": "user", "content": "hi"}],
@@ -80,7 +80,7 @@ def test_anthropic_messages_handler_leaves_native_tools_alone():
         "litellm.llms.anthropic.experimental_pass_through.messages.mcp_handler.anthropic_messages_with_mcp",
         new=AsyncMock(return_value={"routed": True}),
     ) as routed:
-        with pytest.raises(Exception):
+        with pytest.raises(ValueError):
             anthropic_messages_handler(
                 max_tokens=100,
                 messages=[{"role": "user", "content": "hi"}],

tests/test_litellm/llms/azure/videos/test_azure_video_transformation.py

@@ -19,6 +19,7 @@ from litellm.types.videos.main import (
     VideoCreateOptionalRequestParams,
 )
 from litellm.types.router import GenericLiteLLMParams
+from pydantic import ValidationError
 
 
 class TestAzureVideoConfig:
@@ -299,7 +300,7 @@ class TestAzureVideoConfig:
         logging_obj = MagicMock()
 
         # Test that error responses raise exceptions
-        with pytest.raises(Exception):
+        with pytest.raises(ValidationError):
             self.config.transform_video_create_response(
                 model=self.model, raw_response=mock_response, logging_obj=logging_obj
             )

tests/test_litellm/llms/base_llm/search/test_base_search_transformation.py

@@ -270,7 +270,7 @@ async def test_asearch_does_not_leak_server_key_to_caller_api_base(
             new_callable=AsyncMock,
         ) as mock_get,
     ):
-        with pytest.raises(Exception):
+        with pytest.raises(litellm.APIConnectionError):
             await litellm.asearch(
                 query="secrets",
                 search_provider="serper",
@@ -319,7 +319,7 @@ async def test_query_param_key_not_leaked_with_dummy_caller_key(
         "litellm.llms.custom_httpx.http_handler.AsyncHTTPHandler.get",
         fake_get,
     ):
-        with pytest.raises(Exception):
+        with pytest.raises(litellm.APIConnectionError):
             await litellm.asearch(
                 query="secrets",
                 search_provider=provider,

tests/test_litellm/llms/custom_httpx/test_aiohttp_transport.py

@@ -1077,7 +1077,7 @@ async def test_session_closed_retry_does_not_close_concurrent_replacement():
         raise StopAsyncIteration("stop after retry dispatch")
 
     with patch.object(transport, "_make_aiohttp_request", side_effect=fake_make_request):
-        with pytest.raises(Exception):
+        with pytest.raises(StopAsyncIteration):
             await transport.handle_async_request(httpx.Request("GET", "http://example.com"))
 
     try:

tests/test_litellm/llms/github_copilot/test_github_copilot_transformation.py

@@ -866,7 +866,7 @@ class TestGithubCopilotTransformResponse:
         )
         model_response = ModelResponse()
 
-        with pytest.raises(Exception):
+        with pytest.raises(json.JSONDecodeError):
             config.transform_response(
                 model="github_copilot/claude-opus-4.7",
                 raw_response=raw_response,

tests/test_litellm/llms/huggingface/rerank/test_huggingface_rerank_transformation.py

@@ -232,7 +232,7 @@ def test_huggingface_rerank_error_handling(mock_post):
     mock_response.text = "Unauthorized"
     mock_post.return_value = mock_response
 
-    with pytest.raises(Exception):
+    with pytest.raises(litellm.APIConnectionError):
         litellm.rerank(
             model="huggingface/BAAI/bge-reranker-base",
             query="hello",

tests/test_litellm/llms/langflow/chat/test_langflow_chat_transformation.py

@@ -233,7 +233,7 @@ def test_langflow_extra_body_cannot_inject_tweaks_into_run_payload():
         return resp
 
     with patch.object(HTTPHandler, "post", side_effect=fake_post):
-        with pytest.raises(Exception):
+        with pytest.raises(litellm.APIConnectionError):
             litellm.completion(
                 model="langflow/my-flow",
                 messages=[{"role": "user", "content": "hello"}],

tests/test_litellm/llms/vertex_ai/files/test_vertex_ai_files_streaming.py

@@ -40,6 +40,7 @@ from litellm.llms.vertex_ai.files.transformation import (
     _openai_batch_jsonl_entry_to_vertex_rows,
 )
 from litellm.types.llms.openai import CreateFileRequest
+from litellm.llms.vertex_ai.common_utils import VertexAIError
 
 
 def _upload_stream(transformed) -> BaseFileUploadStream:
@@ -561,7 +562,7 @@ class TestStreamingMediaUpload:
 
     async def test_failed_upload_raises(self):
         raw = _make_openai_jsonl_bytes(80)
-        with pytest.raises(Exception):
+        with pytest.raises(VertexAIError):
             await self._run(raw, status=403)
 
     async def test_request_timeout_is_forwarded(self):

tests/test_litellm/llms/xai/test_xai_oauth.py

@@ -556,7 +556,7 @@ def test_get_llm_provider_uses_single_xai_provider(monkeypatch):
 
 
 def test_xai_oauth_alias_is_not_a_provider():
-    with pytest.raises(Exception):
+    with pytest.raises(litellm.BadRequestError):
         get_llm_provider("xai_oauth/grok-4")
 
 

tests/test_litellm/models/test_models.py

@@ -39,6 +39,7 @@ from litellm.models.verification_token import (
     LiteLLM_DeletedVerificationToken,
     LiteLLM_VerificationToken,
 )
+from pydantic import ValidationError
 
 
 class TestBudget:
@@ -421,7 +422,7 @@ class TestBudgetTableFull:
         assert budget.max_budget == 10.0
 
     def test_full_requires_created_at(self):
-        with pytest.raises(Exception):
+        with pytest.raises(ValidationError):
             LiteLLM_BudgetTableFull(budget_id="b1")
 
 
@@ -480,7 +481,7 @@ class TestMCPServerTable:
         assert server.env == {}
 
     def test_mcp_server_requires_transport(self):
-        with pytest.raises(Exception):
+        with pytest.raises(ValidationError):
             LiteLLM_MCPServerTable(server_id="s1")
 
 
@@ -538,7 +539,7 @@ class TestManagedTables:
         assert table.flat_model_file_ids == ["file-abc"]
 
     def test_managed_object_table_requires_purpose(self):
-        with pytest.raises(Exception):
+        with pytest.raises(ValidationError):
             LiteLLM_ManagedObjectTable(
                 unified_object_id="o1", model_object_id="m1", file_object={}
             )

tests/test_litellm/proxy/_experimental/mcp_server/test_mcp_server.py

@@ -4440,7 +4440,7 @@ async def test_call_mcp_tool_logs_failure_via_post_call_failure_hook():
             proxy_logging_mock,
         ),
     ):
-        with pytest.raises(Exception):
+        with pytest.raises(Exception, match="boom"):
             await call_mcp_tool(
                 name="test_server-any_tool",
                 arguments={"x": 1},

tests/test_litellm/proxy/auth/test_route_checks.py

@@ -2809,7 +2809,7 @@ def test_team_update_gate_rejects_without_org_context():
     request.method = "POST"
     request.query_params = {}
 
-    with pytest.raises(Exception):
+    with pytest.raises(Exception, match="Only proxy admin can be used to generate"):
         RouteChecks.non_proxy_admin_allowed_routes_check(
             user_obj=user_obj,
             _user_role=LitellmUserRoles.INTERNAL_USER.value,
@@ -2829,7 +2829,7 @@ def test_team_update_gate_rejects_cross_org_admin_with_resolved_org():
     request.method = "POST"
     request.query_params = {}
 
-    with pytest.raises(Exception):
+    with pytest.raises(Exception, match="Only proxy admin can be used to generate"):
         RouteChecks.non_proxy_admin_allowed_routes_check(
             user_obj=user_obj,
             _user_role=LitellmUserRoles.INTERNAL_USER.value,
@@ -2951,7 +2951,7 @@ def test_patch_team_gate_rejects_regular_internal_user():
     )
     valid_token = UserAPIKeyAuth(user_id="regular-user", user_role=LitellmUserRoles.INTERNAL_USER.value)
 
-    with pytest.raises(Exception):
+    with pytest.raises(Exception, match="Only proxy admin can be used to generate"):
         RouteChecks.non_proxy_admin_allowed_routes_check(
             user_obj=user_obj,
             _user_role=LitellmUserRoles.INTERNAL_USER.value,
@@ -2967,7 +2967,7 @@ def test_patch_team_gate_rejects_cross_org_admin():
     user_obj = _make_org_admin_user("org-1")
     valid_token = UserAPIKeyAuth(user_id="org-admin-user", user_role=LitellmUserRoles.INTERNAL_USER.value)
 
-    with pytest.raises(Exception):
+    with pytest.raises(Exception, match="Only proxy admin can be used to generate"):
         RouteChecks.non_proxy_admin_allowed_routes_check(
             user_obj=user_obj,
             _user_role=LitellmUserRoles.INTERNAL_USER.value,
@@ -2987,7 +2987,7 @@ def test_patch_team_gate_rejects_view_only_admin():
     )
     valid_token = UserAPIKeyAuth(user_id="viewer", user_role=LitellmUserRoles.PROXY_ADMIN_VIEW_ONLY.value)
 
-    with pytest.raises(Exception):
+    with pytest.raises(HTTPException):
         RouteChecks.non_proxy_admin_allowed_routes_check(
             user_obj=user_obj,
             _user_role=LitellmUserRoles.PROXY_ADMIN_VIEW_ONLY.value,

tests/test_litellm/proxy/batches_endpoints/test_endpoints.py

@@ -979,7 +979,7 @@ async def test_create__exception_calls_failure_hook(harness, openai_env_creds):
     )
     harness.litellm_acreate.side_effect = ValueError("provider boom")
 
-    with pytest.raises(Exception):
+    with pytest.raises(ProxyException):
         await call_create(harness)
 
     harness.logging.post_call_failure_hook.assert_called_once()
@@ -1437,7 +1437,7 @@ async def test_retrieve__uses_aretrieve_batch_route_type(retrieve_harness, opena
 async def test_retrieve__exception_calls_failure_hook(retrieve_harness, openai_env_creds):
     retrieve_harness.litellm_aretrieve.side_effect = ValueError("provider boom")
 
-    with pytest.raises(Exception):
+    with pytest.raises(ProxyException):
         await call_retrieve(retrieve_harness, "batch-raw-xyz")
 
     retrieve_harness.logging.post_call_failure_hook.assert_called_once()
@@ -1844,7 +1844,7 @@ async def test_list__uses_alist_batches_route_type(list_harness):
 async def test_list__exception_calls_failure_hook(list_harness):
     list_harness.litellm_alist.side_effect = ValueError("provider boom")
 
-    with pytest.raises(Exception):
+    with pytest.raises(ProxyException):
         await call_list(list_harness)
 
     list_harness.logging.post_call_failure_hook.assert_called_once()
@@ -2233,7 +2233,7 @@ async def test_cancel__uses_acancel_batch_route_type(cancel_harness, openai_env_
 async def test_cancel__exception_calls_failure_hook(cancel_harness, openai_env_creds):
     cancel_harness.litellm_acancel.side_effect = ValueError("provider boom")
 
-    with pytest.raises(Exception):
+    with pytest.raises(ProxyException):
         await call_cancel(cancel_harness, "batch-raw-xyz")
 
     cancel_harness.logging.post_call_failure_hook.assert_called_once()

tests/test_litellm/proxy/db/test_prisma_self_heal.py

@@ -507,7 +507,7 @@ async def test_engine_confirmed_dead_persists_across_failed_heavy_reconnect(
     client._reap_all_zombies = MagicMock()
 
     with patch.dict(os.environ, {"DATABASE_URL": "postgresql://test"}):
-        with pytest.raises(Exception):
+        with pytest.raises(RuntimeError):
             await client._run_reconnect_cycle(timeout_seconds=5.0)
 
     # The flag must STILL be True so the next attempt re-enters the heavy

tests/test_litellm/proxy/guardrails/guardrail_hooks/test_bedrock_guardrails.py

@@ -2113,7 +2113,7 @@ async def test_make_bedrock_api_request_forwards_guardrail_action():
     ):
         mock_post.return_value = mock_bedrock_response
 
-        with pytest.raises(Exception):
+        with pytest.raises(Exception, match="blocked"):
             await guardrail.make_bedrock_api_request(
                 source="INPUT",
                 messages=request_data["messages"],

tests/test_litellm/proxy/guardrails/guardrail_hooks/test_presidio.py

@@ -22,6 +22,7 @@ from litellm.proxy.guardrails.guardrail_hooks.presidio import (
 from litellm.exceptions import GuardrailRaisedException
 from litellm.types.guardrails import LitellmParams, PiiAction, PiiEntityType
 from litellm.types.utils import Choices, Message, ModelResponse
+from litellm.exceptions import BlockedPiiEntityError
 
 
 def _make_mock_session_iterator(
@@ -1345,7 +1346,7 @@ def test_blocking_respects_threshold_filter():
         {"entity_type": PiiEntityType.CREDIT_CARD, "score": 0.95, "start": 0, "end": 4}
     ]
     filtered_high = guardrail.filter_analyze_results_by_score(high_score_results)
-    with pytest.raises(Exception):
+    with pytest.raises(BlockedPiiEntityError):
         guardrail.raise_exception_if_blocked_entities_detected(filtered_high)
 
 

tests/test_litellm/proxy/management_endpoints/scim/test_scim_v2_endpoints.py

@@ -53,6 +53,7 @@ from litellm.types.proxy.management_endpoints.scim_v2 import (
     SCIMUserGroup,
     SCIMUserName,
 )
+from litellm.proxy._types import ProxyException
 
 
 @pytest.mark.asyncio
@@ -3216,7 +3217,7 @@ async def test_delete_user_surfaces_prune_failure_and_keeps_user(mocker):
         AsyncMock(side_effect=Exception("database connection lost")),
     )
 
-    with pytest.raises(Exception):
+    with pytest.raises(ProxyException):
         await delete_user(user_id=user_id)
 
     mock_prisma_client.db.litellm_usertable.delete.assert_not_awaited()

tests/test_litellm/proxy/management_endpoints/test_internal_user_endpoints.py

@@ -668,6 +668,8 @@ def test_validate_sort_params():
     """
     Test that validate_sort_params returns None if sort_by is None
     """
+    from fastapi import HTTPException
+
     from litellm.proxy.management_endpoints.internal_user_endpoints import (
         _validate_sort_params,
     )
@@ -676,7 +678,7 @@ def test_validate_sort_params():
     assert _validate_sort_params(None, "desc") is None
     assert _validate_sort_params("user_id", "asc") == {"user_id": "asc"}
     assert _validate_sort_params("user_id", "desc") == {"user_id": "desc"}
-    with pytest.raises(Exception):
+    with pytest.raises(HTTPException):
         _validate_sort_params("user_id", "invalid")
 
 

tests/test_litellm/proxy/management_helpers/test_team_metadata_validation.py

@@ -22,6 +22,7 @@ from litellm.proxy.management_helpers.team_metadata_validation import (
     run_team_metadata_validation,
     validate_team_metadata_if_configured,
 )
+from pydantic import ValidationError
 
 
 def _registry_with(validator):
@@ -634,7 +635,7 @@ def test_parse_schema_round_trips_fields_in_order():
     ],
 )
 def test_parse_schema_malformed_raises(raw):
-    with pytest.raises(Exception):
+    with pytest.raises(ValidationError):
         parse_team_metadata_schema(raw)
 
 

tests/test_litellm/proxy/pass_through_endpoints/test_pass_through_endpoints.py

@@ -350,7 +350,7 @@ async def test_pass_through_request_failure_handler():
                 mock_user_api_key_dict = MagicMock()
 
                 # Call the function with a target that will trigger an HTTPError
-                with pytest.raises(Exception):
+                with pytest.raises(ProxyException):
                     await pass_through_request(
                         request=mock_request,
                         target="http://test.com",
@@ -1154,7 +1154,7 @@ async def test_pass_through_request_uses_resolved_timeout():
 
             mock_user_api_key_dict = MagicMock()
 
-            with pytest.raises(Exception):
+            with pytest.raises(TypeError):
                 await pass_through_request(
                     request=mock_request,
                     target="http://test.com",

tests/test_litellm/proxy/pass_through_endpoints/test_passthrough_post_call_guardrails.py

@@ -18,6 +18,7 @@ from litellm.integrations.custom_guardrail import (
     CustomGuardrail,
     ModifyResponseException,
 )
+from litellm.proxy._types import ProxyException
 
 _PT_MOD = "litellm.proxy.pass_through_endpoints.pass_through_endpoints"
 _COLLECT = "litellm.proxy.pass_through_endpoints.passthrough_guardrails.PassthroughGuardrailHandler.collect_guardrails"
@@ -251,7 +252,7 @@ class TestPassthroughPostCallGuardrails:
         )
 
         with _common_patches(mock_proxy_logging, mock_response):
-            with pytest.raises(Exception):
+            with pytest.raises(ProxyException):
                 await pass_through_request(
                     request=_make_mock_request(),
                     target="https://example.com/v1/generateContent",

tests/test_litellm/proxy/proxy_server/test_proxy_config.py

@@ -28,6 +28,7 @@ from litellm.proxy.proxy_server import (
 )
 
 from .conftest import normalize
+from pydantic import ValidationError
 
 # ---------------------------------------------------------------------------
 # _is_remote_module_url
@@ -393,7 +394,7 @@ def test_ProxyConfig__load_yaml_file_returns_parsed_dict(tmp_path):
 
 def test_ProxyConfig__load_yaml_file_raises_on_missing_file():
     pc = ProxyConfig()
-    with pytest.raises(Exception):
+    with pytest.raises(Exception, match="Error loading yaml file"):
         pc._load_yaml_file("/no/such/file.yaml")
 
 
@@ -418,7 +419,7 @@ async def test_ProxyConfig__get_config_from_file_loads_yaml(tmp_path):
 @pytest.mark.asyncio
 async def test_ProxyConfig__get_config_from_file_missing_path_raises():
     pc = ProxyConfig()
-    with pytest.raises(Exception):
+    with pytest.raises(Exception, match="Config file not found"):
         await pc._get_config_from_file(config_file_path="/no/such/file.yaml")
 
 
@@ -476,7 +477,7 @@ async def test_ProxyConfig_save_config_invalid_path_raises(monkeypatch):
     monkeypatch.setattr("litellm.proxy.proxy_server.store_model_in_db", False)
     monkeypatch.setattr("litellm.proxy.proxy_server.general_settings", {})
     pc = ProxyConfig()
-    with pytest.raises(Exception):
+    with pytest.raises(FileNotFoundError):
         await pc.save_config({"x": 1})
 
 
@@ -641,7 +642,7 @@ def test_ProxyConfig__get_team_config_returns_match():
 
 def test_ProxyConfig__get_team_config_missing_team_id_raises():
     pc = ProxyConfig()
-    with pytest.raises(Exception):
+    with pytest.raises(Exception, match="team_id missing from team"):
         pc._get_team_config(team_id="t1", all_teams_config=[{"no_id_field": True}])
 
 
@@ -671,7 +672,7 @@ def test_ProxyConfig_load_team_config_no_settings_returns_empty():
     assert out == {}
     # Error-style: a misconfigured team list without team_id raises.
     pc.config = {"litellm_settings": {"default_team_settings": [{"no_id": True}]}}
-    with pytest.raises(Exception):
+    with pytest.raises(Exception, match="team_id missing from team"):
         pc.load_team_config(team_id="anything")
 
 
@@ -698,7 +699,7 @@ def test_ProxyConfig__init_cache_sets_litellm_cache(monkeypatch):
 
 def test_ProxyConfig__init_cache_invalid_params_raises():
     pc = ProxyConfig()
-    with pytest.raises(Exception):
+    with pytest.raises(AttributeError):
         pc._init_cache(cache_params={"type": "this-cache-type-does-not-exist"})
 
 
@@ -765,7 +766,7 @@ async def test_ProxyConfig_get_config_missing_file_raises(monkeypatch):
     monkeypatch.setattr("litellm.proxy.proxy_server.store_model_in_db", False)
     monkeypatch.delenv("LITELLM_CONFIG_BUCKET_NAME", raising=False)
     pc = ProxyConfig()
-    with pytest.raises(Exception):
+    with pytest.raises(Exception, match="Config file not found"):
         await pc.get_config(config_file_path="/no/such/path.yaml")
 
 
@@ -1041,7 +1042,7 @@ def test_ProxyConfig_load_credential_list_returns_items():
 
 def test_ProxyConfig_load_credential_list_invalid_entry_raises():
     pc = ProxyConfig()
-    with pytest.raises(Exception):
+    with pytest.raises(ValidationError):
         pc.load_credential_list({"credential_list": [{"missing_required": True}]})
 
 
@@ -1381,7 +1382,7 @@ async def test_ProxyConfig_load_config_missing_file_raises(monkeypatch):
     monkeypatch.setattr("litellm.proxy.proxy_server.store_model_in_db", False)
     monkeypatch.delenv("LITELLM_CONFIG_BUCKET_NAME", raising=False)
     pc = ProxyConfig()
-    with pytest.raises(Exception):
+    with pytest.raises(Exception, match="Config file not found"):
         await pc.load_config(router=None, config_file_path="/no/file.yaml")
 
 
@@ -1492,7 +1493,7 @@ async def test_ProxyConfig__init_non_llm_configs_empty_config():
 async def test_ProxyConfig__init_non_llm_configs_premium_invalid_worker_registry_raises(monkeypatch):
     monkeypatch.setattr("litellm.proxy.proxy_server.premium_user", True)
     pc = ProxyConfig()
-    with pytest.raises(Exception):
+    with pytest.raises(ValidationError):
         await pc._init_non_llm_configs(
             config={"worker_registry": [{"totally": "invalid"}]},
             config_file_path=None,
@@ -1572,7 +1573,7 @@ async def test_ProxyConfig__init_policy_engine_none_config_noop():
     # None config returns early without raising.
     await pc._init_policy_engine(config=None, prisma_client=None, llm_router=None)
     # Error-style: invalid policies value should raise.
-    with pytest.raises(Exception):
+    with pytest.raises(AttributeError):
         await pc._init_policy_engine(
             config={"policies": "not-a-list"},
             prisma_client=None,
@@ -1601,7 +1602,7 @@ def test_ProxyConfig__load_alerting_settings_noop_when_no_alerting():
 
 def test_ProxyConfig__load_alerting_settings_invalid_alerting_raises():
     pc = ProxyConfig()
-    with pytest.raises(Exception):
+    with pytest.raises(RuntimeError):
         # alerting must be iterable — int triggers an error.
         pc._load_alerting_settings({"alerting": 12345})
 
@@ -1823,7 +1824,7 @@ async def test_ProxyConfig__delete_deployment_invalid_models_raises(monkeypatch)
     fake_router.get_model_ids = MagicMock(return_value=[])
     monkeypatch.setattr("litellm.proxy.proxy_server.llm_router", fake_router)
     pc = ProxyConfig()
-    with pytest.raises(Exception):
+    with pytest.raises(AttributeError):
         # Non-model objects without expected attrs trigger an error.
         await pc._delete_deployment(db_models=[{"not_a_model": True}])
 
@@ -2721,7 +2722,7 @@ async def test_ProxyConfig__update_general_settings_none_input_noop():
     result = await pc._update_general_settings(db_general_settings=None)
     assert result is None
     # Error-style: dict() will fail on non-mapping non-None input.
-    with pytest.raises(Exception):
+    with pytest.raises(TypeError):
         await pc._update_general_settings(db_general_settings=12345)  # type: ignore[arg-type]
 
 
@@ -2743,7 +2744,7 @@ def test_ProxyConfig__update_config_fields_merges_dict():
 
 def test_ProxyConfig__update_config_fields_invalid_param_raises():
     pc = ProxyConfig()
-    with pytest.raises(Exception):
+    with pytest.raises(TypeError):
         # Missing required arg.
         pc._update_config_fields(current_config={}, param_name="general_settings")  # type: ignore[call-arg]
 

tests/test_litellm/proxy/test_proxy_utils.py

@@ -1584,7 +1584,7 @@ async def test_prisma_health_check_failure_redacts_database_credentials(caplog):
     client._report_health_check_failure = AsyncMock()
 
     with caplog.at_level(logging.WARNING, logger="LiteLLM Proxy"):
-        with pytest.raises(Exception):
+        with pytest.raises(Exception, match="could not connect to"):
             await PrismaClient.health_check(client)
 
     emitted = [record.getMessage() for record in caplog.records if record.name == "LiteLLM Proxy"]

tests/test_litellm/test_dashscope_image_generation.py

@@ -17,6 +17,7 @@ from litellm.llms.dashscope.image_generation.transformation import (
 )
 from litellm.types.utils import ImageObject, ImageResponse
 from litellm.utils import get_llm_provider
+from litellm.llms.base_llm.chat.transformation import BaseLLMException
 
 
 # ---------------------------------------------------------------------------
@@ -247,7 +248,7 @@ class TestDashScopeImageGenerationConfig:
             "message": "Size not supported",
         }
 
-        with pytest.raises(Exception):
+        with pytest.raises(BaseLLMException):
             self.cfg.transform_image_generation_response(
                 model="qwen-image-2.0",
                 raw_response=mock_resp,
@@ -268,7 +269,7 @@ class TestDashScopeImageGenerationConfig:
             "message": "Size not supported",
         }
 
-        with pytest.raises(Exception):
+        with pytest.raises(BaseLLMException):
             self.cfg.transform_image_generation_response(
                 model="qwen-image-2.0",
                 raw_response=mock_resp,

tests/test_litellm/test_get_blog_posts.py

@@ -12,6 +12,7 @@ from litellm.litellm_core_utils.get_blog_posts import (
     GetBlogPosts,
     get_blog_posts,
 )
+from xml.etree import ElementTree
 
 SAMPLE_RSS = """\
 <?xml version="1.0" encoding="UTF-8"?>
@@ -71,7 +72,7 @@ def test_parse_rss_to_posts_multiple():
 
 
 def test_parse_rss_to_posts_invalid_xml():
-    with pytest.raises(Exception):
+    with pytest.raises(ElementTree.ParseError):
         GetBlogPosts.parse_rss_to_posts("not xml")
 
 

tests/test_litellm/test_project_tags_pydantic.py

@@ -1,5 +1,6 @@
 import pytest
 from litellm.proxy._types import NewProjectRequest, UpdateProjectRequest
+from pydantic import ValidationError
 
 
 def test_new_project_request_tags():
@@ -21,11 +22,11 @@ def test_update_project_request_tags():
 
 def test_new_project_request_invalid_tags_type():
     # tags must be a list — a string should raise a ValidationError
-    with pytest.raises(Exception):
+    with pytest.raises(ValidationError):
         NewProjectRequest(project_id="test_proj", team_id="team_1", tags="not-a-list")
 
 
 def test_update_project_request_invalid_tags_type():
     # tags must be a list — a string should raise a ValidationError
-    with pytest.raises(Exception):
+    with pytest.raises(ValidationError):
         UpdateProjectRequest(project_id="test_proj", tags="not-a-list")

tests/test_litellm/test_redact_string_in_error_paths.py

@@ -234,7 +234,7 @@ class TestRouterFallbackFailureTracebackRedaction:
                 raise ValueError(f"primary deployment failed api_key={secret}")
             except ValueError as original_exception:
                 with caplog.at_level(logging.DEBUG, logger="LiteLLM Router"):
-                    with pytest.raises(Exception):
+                    with pytest.raises(ValueError):
                         await router.async_function_with_fallbacks_common_utils(
                             e=original_exception,
                             disable_fallbacks=False,

tests/test_litellm/test_retrieve_batch_bedrock_dispatch.py

@@ -23,6 +23,7 @@ import pytest
 sys.path.insert(0, os.path.abspath("../.."))
 
 import litellm  # noqa: E402
+import openai
 
 ASYNC_INVOKE_ARN = "arn:aws:bedrock:us-west-2:123456789012:async-invoke/abc123def456"
 MIJ_ARN = "arn:aws:bedrock:us-west-2:123456789012:model-invocation-job/abc1234567"
@@ -134,7 +135,7 @@ def test_unrelated_bedrock_arn_falls_through_to_provider_config(mock_handlers):
     # Use a plausible-but-unsupported Bedrock ARN family.
     unrelated_arn = "arn:aws:bedrock:us-west-2:123456789012:provisioned-model/xyz"
 
-    with pytest.raises(Exception):
+    with pytest.raises(litellm.BadRequestError):
         # Will raise because no provider_config exists for this path —
         # that's fine, we just need to assert neither bedrock handler ran
         # before the failure.
@@ -152,7 +153,7 @@ def test_non_bedrock_id_skips_bedrock_dispatch_entirely(mock_handlers):
     block — they belong to other providers' retrieve flows."""
     async_invoke, mij, _ = mock_handlers
 
-    with pytest.raises(Exception):
+    with pytest.raises(openai.OpenAIError):
         litellm.retrieve_batch(
             batch_id="batch_abc123",
             custom_llm_provider="openai",

tests/test_litellm/test_router.py

@@ -367,7 +367,7 @@ async def test_arouter_with_tags_and_fallbacks():
         enable_tag_filtering=True,
     )
 
-    with pytest.raises(Exception):
+    with pytest.raises(litellm.InternalServerError):
         response = await router.acompletion(
             model="gpt-3.5-turbo",
             messages=[{"role": "user", "content": "Hello, world!"}],
@@ -5707,7 +5707,7 @@ async def test_team_scoped_model_fallback_cross_team_blocked():
         fallbacks=[{"primary-model": ["fallback-model"]}],
     )
 
-    with pytest.raises(Exception):
+    with pytest.raises(litellm.InternalServerError):
         await router.acompletion(
             model="primary-model",
             messages=[{"role": "user", "content": "Hello"}],

tests/test_litellm/test_router_weighted_failover.py

@@ -391,7 +391,7 @@ async def test_no_failover_when_flag_off():
         # enable_weighted_failover defaults to False
     )
 
-    with pytest.raises(Exception):
+    with pytest.raises(litellm.InternalServerError):
         await router.acompletion(
             model="test-model",
             messages=[{"role": "user", "content": "hi"}],
@@ -515,7 +515,7 @@ async def test_failover_exhausted_raises_original_error_class():
         enable_weighted_failover=True,
     )
 
-    with pytest.raises(Exception):
+    with pytest.raises(litellm.InternalServerError):
         await router.acompletion(
             model="test-model",
             messages=[{"role": "user", "content": "hi"}],
@@ -648,7 +648,7 @@ async def test_failover_skipped_for_non_simple_shuffle():
         enable_weighted_failover=True,
     )
 
-    with pytest.raises(Exception):
+    with pytest.raises(litellm.InternalServerError):
         await router.acompletion(
             model="test-model",
             messages=[{"role": "user", "content": "hi"}],

tests/test_litellm/test_video_generation.py

@@ -151,7 +151,7 @@ class TestVideoGeneration:
             "video_generation_handler",
             side_effect=Exception("API Error"),
         ):
-            with pytest.raises(Exception):
+            with pytest.raises(litellm.APIConnectionError):
                 video_generation(prompt="Test video", model="sora-2")
 
     def test_video_generation_provider_config(self):
@@ -739,7 +739,7 @@ class TestVideoGeneration:
             "video_status_handler",
             side_effect=Exception("API Error"),
         ):
-            with pytest.raises(Exception):
+            with pytest.raises(litellm.APIConnectionError):
                 video_status(video_id="test_video_id", model="sora-2")
 
     def test_video_status_request_transformation(self):

tests/test_litellm/videos/test_main.py

@@ -321,7 +321,7 @@ def test_get_character__mock_response_short_circuits(seams):
 def test_unsupported_provider_raises_without_dispatch(seams):
     seams.get_config.return_value = None
 
-    with pytest.raises(Exception):
+    with pytest.raises(litellm.APIConnectionError):
         videos_main.video_status(video_id=AZURE_VIDEO_ID)
 
     seams.handler.video_status_handler.assert_not_called()

tests/vector_store_tests/test_ragflow_vector_store.py

@@ -16,6 +16,7 @@ from tests.vector_store_tests.base_vector_store_test import BaseVectorStoreTest
 from litellm.llms.ragflow.vector_stores.transformation import RAGFlowVectorStoreConfig
 from litellm.litellm_core_utils.litellm_logging import Logging as LiteLLMLoggingObj
 from litellm.types.vector_stores import VectorStoreCreateOptionalRequestParams
+from litellm.llms.base_llm.chat.transformation import BaseLLMException
 
 
 class TestRAGFlowVectorStore(BaseVectorStoreTest):
@@ -233,7 +234,7 @@ class TestRAGFlowVectorStore(BaseVectorStoreTest):
             "message": "Dataset name 'test-dataset' already exists",
         }
 
-        with pytest.raises(Exception):  # Should raise BaseLLMException
+        with pytest.raises(BaseLLMException):
             config.transform_create_vector_store_response(mock_response)
 
     def test_transform_create_vector_store_response_missing_id(self):