diff --git a/litellm/litellm_core_utils/prompt_templates/factory.py b/litellm/litellm_core_utils/prompt_templates/factory.py
index 77cbe4c9a8e7..da9899cf5e39 100644
--- a/litellm/litellm_core_utils/prompt_templates/factory.py
+++ b/litellm/litellm_core_utils/prompt_templates/factory.py
@@ -40,6 +40,7 @@
 
 from .common_utils import convert_content_list_to_str, is_non_content_values_set
 from .image_handling import convert_url_to_base64
+from pydantic import BaseModel
 
 
 def default_pt(messages):
@@ -1001,6 +1002,8 @@ def infer_protocol_value(
 def _gemini_tool_call_invoke_helper(
     function_call_params: ChatCompletionToolCallFunctionChunk,
 ) -> Optional[VertexFunctionCall]:
+    if isinstance(function_call_params, BaseModel):
+        function_call_params = function_call_params.model_dump()
     name = function_call_params.get("name", "") or ""
     arguments = function_call_params.get("arguments", "")
     if (
@@ -1145,7 +1148,10 @@ def convert_to_gemini_tool_call_result(
                 and prev_tool_call_id
                 and msg_tool_call_id == prev_tool_call_id
             ):
-                name = tool.get("function", {}).get("name", "")
+                function_idf = tool.get("function")
+                if isinstance(function_idf, BaseModel):
+                    function_idf = function_idf.model_dump()
+                name = getattr(function_idf, "get", lambda *_: "")("name", "")
 
     if not name:
         raise Exception(
diff --git a/litellm/llms/custom_httpx/llm_http_handler.py b/litellm/llms/custom_httpx/llm_http_handler.py
index d404077a5b62..2d0cef9be5b0 100644
--- a/litellm/llms/custom_httpx/llm_http_handler.py
+++ b/litellm/llms/custom_httpx/llm_http_handler.py
@@ -79,6 +79,7 @@
     ModelResponse,
     ProviderConfigManager,
 )
+from pydantic import BaseModel
 
 if TYPE_CHECKING:
     from litellm.litellm_core_utils.litellm_logging import Logging as _LiteLLMLoggingObj
@@ -108,6 +109,16 @@ async def _make_common_async_call(
             provider_config.max_retry_on_unprocessable_entity_error
         )
 
+        for message in data.get("messages", []):
+            if message.get("role") == "assistant" and message.get("tool_calls"):
+                for tool in message["tool_calls"]:
+                    fn = tool.get("function")
+                    if fn:
+                        tool["function"] = (
+                            fn.model_dump() if isinstance(fn, BaseModel)
+                            else getattr(fn, "__dict__", fn)
+                        )
+
         response: Optional[httpx.Response] = None
         for i in range(max(max_retry_on_unprocessable_entity_error, 1)):
             try: