diff --git a/openjudge/models/openai_chat_model.py b/openjudge/models/openai_chat_model.py
index 1d5294520..4a9bc97f4 100644
--- a/openjudge/models/openai_chat_model.py
+++ b/openjudge/models/openai_chat_model.py
@@ -1,5 +1,6 @@
 # -*- coding: utf-8 -*-
 """OpenAI Client."""
+
 import copy
 import os
 from typing import Any, AsyncGenerator, Callable, Dict, Literal, Type
@@ -27,6 +28,16 @@ def _format_audio_data_for_qwen_omni(messages: list[dict | ChatMessage]) -> list
     """
     format_data = []
     for msg in messages:
+        msg_dict = msg.to_dict() if isinstance(msg, ChatMessage) else msg
+        if isinstance(msg_dict.get("content"), list):
+            for block in msg_dict["content"]:
+                if (
+                    isinstance(block, dict)
+                    and "input_audio" in block
+                    and isinstance(block["input_audio"].get("data"), str)
+                ):
+                    if not block["input_audio"]["data"].startswith("http"):
+                        block["input_audio"]["data"] = "data:;base64," + block["input_audio"]["data"]
         try:
             msg_copy = copy.deepcopy(msg)
             msg_dict = msg_copy.to_dict() if isinstance(msg_copy, ChatMessage) else msg_copy
@@ -58,6 +69,8 @@ def __init__(
         reasoning_effort: Literal["low", "medium", "high"] | None = None,
         organization: str | None = None,
         client_args: Dict[str, Any] | None = None,
+        max_retries: int | None = None,
+        timeout: float | None = None,
         **kwargs: Any,
     ) -> None:
         """Initialize the openai client.
@@ -94,6 +107,11 @@ def __init__(
 
         if organization:
             client_args["organization"] = organization
+        if max_retries is not None:
+            client_args["max_retries"] = max_retries
+
+        if timeout is not None:
+            client_args["timeout"] = timeout
 
         self.client = AsyncOpenAI(**client_args)
 
@@ -289,7 +307,10 @@ def _handle_non_streaming_response(
                     parsed_response.parsed.update(callback_result)
             except Exception as e:
                 # Log the exception but don't fail the entire operation
-                logger.warning(f"Callback function raised an exception: {type(e).__name__}: {e}", exc_info=True)
+                logger.warning(
+                    f"Callback function raised an exception: {type(e).__name__}: {e}",
+                    exc_info=True,
+                )
 
         return parsed_response
 
@@ -386,6 +407,9 @@ async def _handle_streaming_response(
                             final_response.parsed = final_response.parsed or {}
                             final_response.parsed.update(callback_result)
                     except Exception as e:
-                        logger.warning(f"Callback function raised an exception: {type(e).__name__}: {e}", exc_info=True)
+                        logger.warning(
+                            f"Callback function raised an exception: {type(e).__name__}: {e}",
+                            exc_info=True,
+                        )
 
                 yield final_response
diff --git a/tests/models/test_openai_chat_model.py b/tests/models/test_openai_chat_model.py
index c3073f3be..237f15247 100644
--- a/tests/models/test_openai_chat_model.py
+++ b/tests/models/test_openai_chat_model.py
@@ -1,5 +1,6 @@
 # -*- coding: utf-8 -*-
 """Unit tests for OpenAIChatModel."""
+
 import asyncio
 from unittest.mock import AsyncMock, patch
 
@@ -298,6 +299,54 @@ def test_qwen_omni_audio_formatting(self):
             "data:;base64,",
         )
 
+    @pytest.mark.parametrize(
+        "init_kwargs, expected_in_call, not_expected_in_call",
+        [
+            (
+                {"max_retries": 5, "timeout": 120.0},
+                {"max_retries": 5, "timeout": 120.0},
+                [],
+            ),
+            ({}, {}, ["max_retries", "timeout"]),
+            ({"max_retries": None, "timeout": None}, {}, ["max_retries", "timeout"]),
+            ({"max_retries": 3}, {"max_retries": 3}, ["timeout"]),
+            ({"timeout": 30.0}, {"timeout": 30.0}, ["max_retries"]),
+        ],
+        ids=[
+            "with_retries_and_timeout",
+            "defaults",
+            "with_none_values",
+            "with_retries_only",
+            "with_timeout_only",
+        ],
+    )
+    @patch("openjudge.models.openai_chat_model.AsyncOpenAI")
+    def test_client_initialization_with_retries_and_timeout(
+        self,
+        mock_async_openai,
+        init_kwargs,
+        expected_in_call,
+        not_expected_in_call,
+    ):
+        """Test that max_retries and timeout parameters are passed to AsyncOpenAI client correctly."""
+        OpenAIChatModel(
+            model="gpt-4",
+            api_key="test-key",
+            **init_kwargs,
+        )
+
+        mock_async_openai.assert_called_once()
+        call_kwargs = mock_async_openai.call_args[1]
+
+        assert call_kwargs["api_key"] == "test-key"
+
+        for key, value in expected_in_call.items():
+            assert key in call_kwargs
+            assert call_kwargs[key] == value
+
+        for key in not_expected_in_call:
+            assert key not in call_kwargs
+
 
 if __name__ == "__main__":
     pytest.main([__file__])