diff --git a/api/utils/api_utils.py b/api/utils/api_utils.py
index e9fd8af972..d8f052577f 100644
--- a/api/utils/api_utils.py
+++ b/api/utils/api_utils.py
@@ -390,7 +390,7 @@ def get_parser_config(chunk_method, parser_config):
         "one": None,
         "knowledge_graph": {
             "chunk_token_num": 8192,
-            "delimiter": r"\n",
+            "delimiter": "\n",
             "entity_types": ["organization", "person", "location", "event", "time"],
             "raptor": {"use_raptor": False},
             "graphrag": {"use_graphrag": False},
diff --git a/api/utils/validation_utils.py b/api/utils/validation_utils.py
index 4a54756e73..0f7b22f970 100644
--- a/api/utils/validation_utils.py
+++ b/api/utils/validation_utils.py
@@ -425,7 +425,7 @@ class ParentChildConfig(Base):
     """Dataset parser configuration for parent-child chunking."""
 
     use_parent_child: Annotated[bool, Field(default=False)]
-    children_delimiter: Annotated[str, Field(default=r"\n", min_length=1)]
+    children_delimiter: Annotated[str, Field(default="\n", min_length=1)]
 
 
 class AutoMetadataField(Base):
@@ -453,7 +453,7 @@ class ParserConfig(Base):
     auto_keywords: Annotated[int, Field(default=0, ge=0, le=32)]
     auto_questions: Annotated[int, Field(default=0, ge=0, le=10)]
     chunk_token_num: Annotated[int, Field(default=512, ge=1, le=2048)]
-    delimiter: Annotated[str, Field(default=r"\n", min_length=1)]
+    delimiter: Annotated[str, Field(default="\n", min_length=1)]
     graphrag: Annotated[GraphragConfig, Field(default_factory=lambda: GraphragConfig(use_graphrag=False))]
     html4excel: Annotated[bool, Field(default=False)]
     layout_recognize: Annotated[str, Field(default="DeepDOC")]
diff --git a/test/unit_test/api/utils/test_default_delimiter.py b/test/unit_test/api/utils/test_default_delimiter.py
new file mode 100644
index 0000000000..b0d0ff958f
--- /dev/null
+++ b/test/unit_test/api/utils/test_default_delimiter.py
@@ -0,0 +1,50 @@
+#
+#  Copyright 2025 The InfiniFlow Authors. All Rights Reserved.
+#
+#  Licensed under the Apache License, Version 2.0 (the "License");
+#  you may not use this file except in compliance with the License.
+#  You may obtain a copy of the License at
+#
+#      http://www.apache.org/licenses/LICENSE-2.0
+#
+#  Unless required by applicable law or agreed to in writing, software
+#  distributed under the License is distributed on an "AS IS" BASIS,
+#  WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+#  See the License for the specific language governing permissions and
+#  limitations under the License.
+#
+
+"""The API default delimiters must be an actual newline, not the raw 2-char
+escape ``r"\\n"``.
+
+A default stored as ``r"\\n"`` reaches ``rag.nlp.delim.parse_delimiter_field`` as
+a backslash followed by the letter ``n``. The canonical parser honours bare
+characters, so it then splits documents on every ``n`` (``"information"`` ->
+``"i"``, ``"formatio"``) and on the stray backslash. An actual newline parses to
+a single newline delimiter. The frontend already sends a real newline and the
+``naive`` default already uses one; these guard the ``knowledge_graph`` and
+``ParserConfig`` / ``ParentChildConfig`` defaults that did not.
+"""
+
+from api.utils.api_utils import get_parser_config
+from api.utils.validation_utils import ParentChildConfig, ParserConfig
+from rag.nlp.delim import parse_delimiter_field
+
+
+def _assert_real_newline(delimiter):
+    assert delimiter == "\n", f"default delimiter should be a real newline, got {delimiter!r}"
+    # A real newline parses to a single newline delimiter. The raw escape would
+    # instead parse to ["\\", "n"] and shred the document on the letter n.
+    assert parse_delimiter_field(delimiter) == ["\n"]
+
+
+def test_parser_config_delimiter_default_is_real_newline():
+    _assert_real_newline(ParserConfig().delimiter)
+
+
+def test_parent_child_children_delimiter_default_is_real_newline():
+    _assert_real_newline(ParentChildConfig().children_delimiter)
+
+
+def test_knowledge_graph_delimiter_default_is_real_newline():
+    _assert_real_newline(get_parser_config("knowledge_graph", None)["delimiter"])