diff --git a/modelx/serialize/serializer_6.py b/modelx/serialize/serializer_6.py
index 8bfc10ca..e0efd6d2 100644
--- a/modelx/serialize/serializer_6.py
+++ b/modelx/serialize/serializer_6.py
@@ -36,6 +36,12 @@
     IOSpecUnpickler, ModelUnpickler,
     IOSpecPickler, ModelPickler)
 
+PSEUDO_PYTHON_HEADER = """\
+# modelx: pseudo-python
+# This file is part of a modelx model.
+# It can be imported as a Python module, but functions defined herein
+# are model formulas and may not be executable as standard Python."""
+
 
 class TupleID(tuple):
 
@@ -455,7 +461,7 @@ def __init__(self, writer,
         )
 
     def encode(self):
-        lines = []
+        lines = [PSEUDO_PYTHON_HEADER]
         if self.model.doc is not None:
             lines.append("\"\"\"" + self.model.doc + "\"\"\"")
 
@@ -513,7 +519,7 @@ def __init__(self, writer, target, srcpath=None):
 
     def encode(self):
 
-        lines = []
+        lines = [PSEUDO_PYTHON_HEADER]
         if self.space.doc is not None:
             lines.append("\"\"\"" + self.space.doc + "\"\"\"")
 
@@ -1037,12 +1043,11 @@ class DocstringParser(BaseNodeParser):
     """Docstring at module level"""
     @classmethod
     def condition(cls, stmt: StatementTokens):
-        # stmt has 1 element that is STRING and starts at line 1.
+        # stmt has 1 element that is STRING.
         if stmt.section == "DEFAULT" and len(stmt) == 1:
             elm = stmt[0]
             if elm.type == tokenize.STRING:
-                if elm.start[0] == 1:
-                    return True
+                return True
 
         return False
 
diff --git a/modelx/tests/serialize/test_serialize.py b/modelx/tests/serialize/test_serialize.py
index a3cfe2b2..ae821c21 100644
--- a/modelx/tests/serialize/test_serialize.py
+++ b/modelx/tests/serialize/test_serialize.py
@@ -280,3 +280,30 @@ def test_false_value(tmp_path, write_method):
     s.a = False
     getattr(mx, write_method)(m, tmp_path / "model")
     m2 = mx.read_model(tmp_path / "model")
+
+
+def test_pseudo_python_header(tmp_path):
+    from modelx.serialize.serializer_6 import PSEUDO_PYTHON_HEADER
+    m = mx.new_model("HeaderTest")
+    s = m.new_space("Space1")
+
+    @mx.defcells
+    def foo(x):
+        return x
+
+    mx.write_model(m, tmp_path / "model")
+
+    # Check model __init__.py
+    model_init = (tmp_path / "model" / "__init__.py").read_text()
+    assert model_init.startswith(PSEUDO_PYTHON_HEADER)
+
+    # Check space __init__.py
+    space_init = (tmp_path / "model" / "Space1" / "__init__.py").read_text()
+    assert space_init.startswith(PSEUDO_PYTHON_HEADER)
+
+    # Verify the model can be read back correctly
+    m2 = mx.read_model(tmp_path / "model")
+    assert m2.name == "HeaderTest"
+    assert m2.Space1.foo(3) == 3
+    m2.close()
+    m.close()