diff --git a/core/plugin/builtin_plugins/kira-ai/tags.py b/core/plugin/builtin_plugins/kira-ai/tags.py
index 0a0ac217..f6a70b3a 100644
--- a/core/plugin/builtin_plugins/kira-ai/tags.py
+++ b/core/plugin/builtin_plugins/kira-ai/tags.py
@@ -81,9 +81,8 @@ async def handle(self, value: str, **kwargs) -> list[BaseMessageElement]:
                 if not ref_file.is_file():
                     message_logger.warning(f"Image reference not found: {ref_file}, skipped")
                     continue
-                img_extension = ref_file.suffix.lstrip(".")
                 bs64 = await image_to_base64(str(ref_file))
-                ref_images.append(Image(image=bs64, name=p, mime=f"image/{img_extension}"))
+                ref_images.append(Image(image=bs64, name=p))
             try:
                 image_client = self.ctx.provider_mgr.get_default_image()
             except ValueError as exc:
@@ -207,13 +206,12 @@ async def handle(self, value: str, **kwargs) -> list[BaseMessageElement]:
             else:
                 ref_file = get_data_path() / ref_img_path
             if ref_file.is_file():
-                img_extension = ref_file.suffix.lstrip(".")
                 bs64 = await image_to_base64(str(ref_file))
                 image_client = self.ctx.provider_mgr.get_default_image()
                 if not image_client:
                     message_logger.error("Failed to get image client, please set default image model in Configuration")
                     return []
-                img_res = await image_to_image(image_client, value, image=Image(image=bs64, name=ref_img_path, mime=f"image/{img_extension}"))
+                img_res = await image_to_image(image_client, value, image=Image(image=bs64, name=ref_img_path))
                 if img_res:
                     return [img_res]
                 message_logger.warning("Invalid selfie image result")
diff --git a/core/provider/src/openai/model_clients.py b/core/provider/src/openai/model_clients.py
index 4cc4ea67..7ca1f813 100644
--- a/core/provider/src/openai/model_clients.py
+++ b/core/provider/src/openai/model_clients.py
@@ -96,6 +96,8 @@ def _image_file_from_data_url(
 
         mime_type, *parameters = header.removeprefix("data:").split(";")
         mime_type = mime_type.lower()
+        if mime_type == "image/jpg":
+            mime_type = "image/jpeg"
         if "base64" not in {parameter.lower() for parameter in parameters}:
             raise ValueError("Image Data URL must contain Base64 data")
 
diff --git a/tests/test_openai_image_client.py b/tests/test_openai_image_client.py
index b41676f5..a86936c1 100644
--- a/tests/test_openai_image_client.py
+++ b/tests/test_openai_image_client.py
@@ -167,3 +167,30 @@ async def test_image_edit_requires_input_image():
         await client.image_to_image("edit it", [])
 
     edit.assert_not_awaited()
+
+
+@pytest.mark.asyncio
+@pytest.mark.parametrize("mime", ["image/jpg", "image/JPG", "image/jpeg"])
+@pytest.mark.parametrize("source", ["data_url", "base64"])
+async def test_image_edit_normalizes_jpeg_mime(mime, source):
+    jpeg_bytes = b"\xff\xd8\xff\xe0jpeg"
+    encoded = base64.b64encode(jpeg_bytes).decode()
+    image = (
+        Image(image=f"data:{mime};base64,{encoded}")
+        if source == "data_url"
+        else Image(image=encoded, mime=mime)
+    )
+    edit = AsyncMock(
+        return_value=SimpleNamespace(
+            data=[SimpleNamespace(url=None, b64_json=PNG_BASE64)]
+        )
+    )
+    client = _configured_client(SimpleNamespace(edit=edit))
+
+    result = await client.image_to_image("edit it", image)
+
+    assert result.image == PNG_BASE64
+    edit.assert_awaited_once()
+    assert edit.await_args.kwargs["image"] == (
+        "image_0.jpg", jpeg_bytes, "image/jpeg"
+    )