diff --git a/litellm/exceptions.py b/litellm/exceptions.py
index fd328fdfed6..ceb484015db 100644
--- a/litellm/exceptions.py
+++ b/litellm/exceptions.py
@@ -425,9 +425,9 @@ class RateLimitError(openai.RateLimitError):  # type: ignore
 
     def __init__(
         self,
-        message,
-        llm_provider,
-        model,
+        message: str,
+        llm_provider: str,
+        model: Optional[str],
         response: Optional[httpx.Response] = None,
         litellm_debug_info: Optional[str] = None,
         max_retries: Optional[int] = None,
@@ -435,7 +435,7 @@ class RateLimitError(openai.RateLimitError):  # type: ignore
         category: Union[str, RateLimitErrorCategory] = (RateLimitErrorCategory.VENDOR_RATE_LIMIT),
         rate_limit_type: Optional[Union[str, RateLimitType]] = None,
         headers: Optional[Dict[str, str]] = None,
-        detail: Any = None,
+        detail: object = None,
     ):
         self.status_code = 429
         self.message = "litellm.RateLimitError: {}".format(message)
@@ -509,9 +509,9 @@ class RateLimitError(openai.RateLimitError):  # type: ignore
 class InsufficientQuotaError(RateLimitError):
     def __init__(
         self,
-        message,
-        llm_provider,
-        model,
+        message: str,
+        llm_provider: str,
+        model: Optional[str],
         response: Optional[httpx.Response] = None,
         litellm_debug_info: Optional[str] = None,
         max_retries: Optional[int] = None,
@@ -519,7 +519,7 @@ class InsufficientQuotaError(RateLimitError):
         category: Union[str, RateLimitErrorCategory] = (RateLimitErrorCategory.VENDOR_RATE_LIMIT),
         rate_limit_type: Optional[Union[str, RateLimitType]] = None,
         headers: Optional[Dict[str, str]] = None,
-        detail: Any = None,
+        detail: object = None,
     ):
         super().__init__(
             message=message,
diff --git a/litellm/litellm_core_utils/exception_mapping_utils.py b/litellm/litellm_core_utils/exception_mapping_utils.py
index e85f6de6a6c..92a0cba1173 100644
--- a/litellm/litellm_core_utils/exception_mapping_utils.py
+++ b/litellm/litellm_core_utils/exception_mapping_utils.py
@@ -4,6 +4,7 @@ import traceback
 from typing import Any, Optional, Protocol, cast
 
 import httpx
+from pydantic import TypeAdapter
 
 import litellm
 from litellm._logging import _ENABLE_SECRET_REDACTION, _redact_string, verbose_logger
@@ -84,9 +85,6 @@ class ExceptionCheckers:
         Returns:
             True if the error indicates insufficient quota, False otherwise
         """
-        if not isinstance(error_str, str):
-            return False
-
         return "insufficient_quota" in error_str.lower()
 
     @staticmethod
@@ -201,6 +199,9 @@ def _get_body_error_code(error_str: str) -> int | None:
         return None
 
 
+_ERROR_BODY_ADAPTER: TypeAdapter[dict[object, object]] = TypeAdapter(dict[object, object])
+
+
 def _body_has_insufficient_quota_code(error_obj: object) -> bool:
     """True only when the provider's STRUCTURED error body carries
     code or type == "insufficient_quota".
@@ -211,26 +212,25 @@ def _body_has_insufficient_quota_code(error_obj: object) -> bool:
     missing or unparseable body returns False, so callers degrade to the plain
     retryable RateLimitError rather than promoting on a free-text match.
     """
-    try:
-        if error_obj is None:
-            return False
-        if getattr(error_obj, "code", None) == "insufficient_quota":
-            return True
-        if getattr(error_obj, "type", None) == "insufficient_quota":
-            return True
-        body = getattr(error_obj, "body", None)
-        if not isinstance(body, dict):
-            return False
-        nested = body.get("error")
-        candidates = (
-            body.get("code"),
-            body.get("type"),
-            nested.get("code") if isinstance(nested, dict) else None,
-            nested.get("type") if isinstance(nested, dict) else None,
-        )
-        return "insufficient_quota" in candidates
-    except Exception:
+    if error_obj is None:
         return False
+    if getattr(error_obj, "code", None) == "insufficient_quota":
+        return True
+    if getattr(error_obj, "type", None) == "insufficient_quota":
+        return True
+    raw_body = getattr(error_obj, "body", None)
+    if not isinstance(raw_body, dict):
+        return False
+    body = _ERROR_BODY_ADAPTER.validate_python(raw_body)
+    raw_nested = body.get("error")
+    nested = _ERROR_BODY_ADAPTER.validate_python(raw_nested) if isinstance(raw_nested, dict) else None
+    candidates = (
+        body.get("code"),
+        body.get("type"),
+        nested.get("code") if nested is not None else None,
+        nested.get("type") if nested is not None else None,
+    )
+    return "insufficient_quota" in candidates
 
 
 def _get_response_headers(original_exception: Exception) -> Optional[httpx.Headers]:
diff --git a/tests/test_litellm/litellm_core_utils/test_exception_mapping_utils.py b/tests/test_litellm/litellm_core_utils/test_exception_mapping_utils.py
index fa5e48f953c..0a58045271c 100644
--- a/tests/test_litellm/litellm_core_utils/test_exception_mapping_utils.py
+++ b/tests/test_litellm/litellm_core_utils/test_exception_mapping_utils.py
@@ -749,9 +749,6 @@ class TestIsErrorStrInsufficientQuota:
             is False
         )
 
-    def test_non_string_returns_false(self):
-        assert ExceptionCheckers.is_error_str_insufficient_quota(None) is False  # type: ignore
-
 
 def test_openai_insufficient_quota_maps_to_insufficient_quota_error():
     """