diff --git a/docs/source/user-guide/latest/expressions.md b/docs/source/user-guide/latest/expressions.md
index 2ba71e93c4d..33683e37383 100644
--- a/docs/source/user-guide/latest/expressions.md
+++ b/docs/source/user-guide/latest/expressions.md
@@ -164,7 +164,7 @@ The tables below list every Spark built-in expression with its current status.
 | `array_repeat` | ✅ | Native |  |
 | `array_union` | ✅ | Native | NaN/signed-zero handling may differ ([details](compatibility/floating-point.md)) |
 | `arrays_overlap` | ✅ | Native |  |
-| `arrays_zip` | ✅ | Native |  |
+| `arrays_zip` | ✅ | Hybrid | Arrays of maps route through the JVM codegen dispatcher; scalar/array/struct elements run natively. NullType elements fall back to Spark |
 | `element_at` | ✅ | Native |  |
 | `flatten` | ✅ | Native | Binary/struct/map elements fall back |
 | `get` | ✅ | — |  |
diff --git a/spark/src/main/scala/org/apache/comet/serde/arrays.scala b/spark/src/main/scala/org/apache/comet/serde/arrays.scala
index 63860c629cc..69638346b8b 100644
--- a/spark/src/main/scala/org/apache/comet/serde/arrays.scala
+++ b/spark/src/main/scala/org/apache/comet/serde/arrays.scala
@@ -814,10 +814,14 @@ object CometArrayPosition extends CometExpressionSerde[ArrayPosition] with Array
   }
 }
 
-object CometArraysZip extends CometExpressionSerde[ArraysZip] {
+object CometArraysZip extends CometExpressionSerde[ArraysZip] with CodegenDispatchFallback {
 
+  // MapType elements have no native impl; they route through the JVM codegen dispatcher.
+  // This reason surfaces only when the dispatcher is disabled or refuses the tree.
   override def getUnsupportedReasons(): Seq[String] = Seq(
-    "Not all input data types are supported; falls back to Spark for unsupported types")
+    "MapType element arrays are not supported natively and are handled via JVM codegen dispatch; " +
+      "this fallback only applies when the dispatcher is disabled. NullType elements remain " +
+      "unsupported (the dispatcher's isSupportedDataType does not admit NullType).")
 
   private def isTypeSupported(dt: DataType): Boolean = {
     import DataTypes._
diff --git a/spark/src/test/resources/sql-tests/expressions/array/arrays_zip.sql b/spark/src/test/resources/sql-tests/expressions/array/arrays_zip.sql
index e62e3842e8f..265f5231223 100644
--- a/spark/src/test/resources/sql-tests/expressions/array/arrays_zip.sql
+++ b/spark/src/test/resources/sql-tests/expressions/array/arrays_zip.sql
@@ -218,12 +218,18 @@ SELECT arrays_zip(array(struct(1, 2, 3), struct(2, 3, 4)));
 -- query
 -- SELECT arrays_zip(array(struct(1, 2, 3), struct(2, 3, 4), struct(null, null, null)));
 
--- Arrays of maps
--- FIXME: COMET: map is not supported, unsupported arguments for CreateArray, unsupported arguments for ArraysZip
--- +------------------------------------------------------------------+
--- |arrays_zip(array(map(1.0, 2, 3.0, 4)), array(map(1.0, 2, 3.0, 4)))|
--- +------------------------------------------------------------------+
--- |[{{1.0 -> 2, 3.0 -> 4}, {1.0 -> 2, 3.0 -> 4}}]                    |
--- +------------------------------------------------------------------+
--- query
--- SELECT arrays_zip(array(map(1.0, '2', 3.0, '4')), array(map(1.0, '2', 3.0, '4')));
+-- arrays_zip over arrays of maps: MapType elements have no native impl; the projection stays
+-- in Comet via the JVM codegen dispatcher. Input is built from a table column so the
+-- expression is not constant-folded before reaching Comet.
+statement
+CREATE TABLE arrays_zip_map_tbl using parquet AS SELECT 1 AS k1, 2 AS v1, 3 AS k2, 4 AS v2;
+
+-- positive: arrays of maps must execute via dispatch, not whole-projection Spark fallback
+query
+SELECT arrays_zip(array(map(k1, v1), map(k2, v2)), array(map('x', v2), map('y', v1)))
+FROM arrays_zip_map_tbl
+
+-- mixed projection: map-element dispatch + scalar native path in the same projection
+query
+SELECT arrays_zip(array(map(k1, v1)), array(k1, k2)), v1 + v2
+FROM arrays_zip_map_tbl