diff --git a/litellm/responses/litellm_completion_transformation/transformation.py b/litellm/responses/litellm_completion_transformation/transformation.py
index 8f5cd72cbcd..b8e60d40e29 100644
--- a/litellm/responses/litellm_completion_transformation/transformation.py
+++ b/litellm/responses/litellm_completion_transformation/transformation.py
@@ -2259,6 +2259,9 @@ class LiteLLMCompletionResponsesConfig:
             file_dict["file_data"] = item["file_data"]
         if item.get("filename"):
             file_dict["filename"] = item["filename"]
+        media_format: Final = item.get("format")
+        if media_format is not None:
+            file_dict["format"] = media_format
         bucket_owner: Final = item.get("bucket_owner")
         if bucket_owner is not None and _is_s3_url(file_id or item.get("file_data")):
             file_dict["bucket_owner"] = bucket_owner
@@ -2283,6 +2286,9 @@ class LiteLLMCompletionResponsesConfig:
             if bucket_owner is not None and _is_s3_url(url)
             else ChatCompletionImageUrlObject(url=url, detail=detail)
         )
+        media_format: Final = item.get("format")
+        if media_format is not None:
+            image_url_obj["format"] = media_format
 
         return ChatCompletionImageObject(type="image_url", image_url=image_url_obj)
 
diff --git a/tests/unit/responses/litellm_completion_transformation/test_litellm_completion_responses.py b/tests/unit/responses/litellm_completion_transformation/test_litellm_completion_responses.py
index c3a74465139..c17946a2f6b 100644
--- a/tests/unit/responses/litellm_completion_transformation/test_litellm_completion_responses.py
+++ b/tests/unit/responses/litellm_completion_transformation/test_litellm_completion_responses.py
@@ -44,6 +44,68 @@ from litellm.types.utils import (
 
 
 class TestLiteLLMCompletionResponsesConfig:
+    @pytest.mark.parametrize("part_type", ["input_image", "input_file"])
+    @pytest.mark.parametrize(
+        ("key", "media_format", "expected_format"),
+        [("key", "image/png", "png"), ("key.jpeg", "image/png", "png"), ("key.png", None, "png")],
+    )
+    def test_explicit_responses_media_format_reaches_the_converse_s3_image(
+        self, part_type: str, key: str, media_format: str | None, expected_format: str
+    ) -> None:
+        uri: Final = f"s3://bucket/{key}"
+        messages: Final = LiteLLMCompletionResponsesConfig.transform_responses_api_input_to_messages(
+            input=[
+                {
+                    "role": "user",
+                    "content": [
+                        {
+                            "type": part_type,
+                            "image_url" if part_type == "input_image" else "file_id": uri,
+                            **({"format": media_format} if media_format is not None else {}),
+                        }
+                    ],
+                }
+            ],
+            responses_api_request={},
+        )
+        request: Final = litellm.AmazonConverseConfig()._transform_request(
+            model="us.amazon.nova-2-lite-v1:0",
+            messages=messages,
+            optional_params={},
+            litellm_params={},
+            headers={},
+        )
+
+        assert request["messages"] == [
+            {
+                "role": "user",
+                "content": [{"image": {"format": expected_format, "source": {"s3Location": {"uri": uri}}}}],
+            }
+        ]
+
+    @pytest.mark.parametrize("part_type", ["input_image", "input_file"])
+    def test_extensionless_responses_s3_media_without_format_is_a_bad_request(self, part_type: str) -> None:
+        messages: Final = LiteLLMCompletionResponsesConfig.transform_responses_api_input_to_messages(
+            input=[
+                {
+                    "role": "user",
+                    "content": [
+                        {"type": part_type, "image_url" if part_type == "input_image" else "file_id": "s3://bucket/key"}
+                    ],
+                }
+            ],
+            responses_api_request={},
+        )
+
+        with pytest.raises(litellm.BadRequestError, match="Cannot tell the media type"):
+            litellm.AmazonConverseConfig()._transform_request(
+                model="us.amazon.nova-2-lite-v1:0",
+                messages=messages,
+                optional_params={},
+                litellm_params={},
+                headers={},
+            )
+
     def test_transform_input_file_item_to_file_item_with_file_id(self):
         """Test transformation of input_file item with file_id to Chat Completion file format"""
         # Setup