diff --git a/examples/benchmark/tasks_info.py b/examples/benchmark/tasks_info.py
index 9c56172..64b9c1e 100644
--- a/examples/benchmark/tasks_info.py
+++ b/examples/benchmark/tasks_info.py
@@ -12,15 +12,16 @@ class AvailableTask(StrEnum):
     """Available tasks for training."""
 
     # Complex tasks
-    PREPARE_MEAL = TaskType.PREPARE_MEAL
-    RELAX_ON_SOFA = TaskType.RELAX_ON_SOFA
-    READ_BOOK_IN_BED = TaskType.READ_BOOK_IN_BED
-    SETUP_BATH = TaskType.SETUP_BATH
-    MULTI_TASK = "MultiTask"
     CLEAN_UP_KITCHEN = TaskType.CLEAN_UP_KITCHEN
     CLEAN_UP_LIVING_ROOM = TaskType.CLEAN_UP_LIVING_ROOM
     CLEAN_UP_BEDROOM = TaskType.CLEAN_UP_BEDROOM
     CLEAN_UP_BATHROOM = TaskType.CLEAN_UP_BATHROOM
+    MULTI_TASK_4 = "MultiTask4"
+    MULTI_TASK_8 = "MultiTask8"
+    PREPARE_MEAL = TaskType.PREPARE_MEAL
+    RELAX_ON_SOFA = TaskType.RELAX_ON_SOFA
+    READ_BOOK_IN_BED = TaskType.READ_BOOK_IN_BED
+    SETUP_BATH = TaskType.SETUP_BATH
 
     # Gradual tasks
     # 1 item
@@ -457,7 +458,17 @@ def keep_only_n_scenes(task_blueprint_config: dict[str, Any], nb_scenes: int) ->
 def get_task_blueprint_config(task: AvailableTask, nb_scenes: int) -> list[dict[str, Any]]:
     """Return the scenes for the task."""
     match task:
-        case AvailableTask.MULTI_TASK:
+        case AvailableTask.MULTI_TASK_4:
+            return [
+                keep_only_n_scenes(task_blueprints_configs[task], nb_scenes)
+                for task in (
+                    AvailableTask.CLEAN_UP_KITCHEN,
+                    AvailableTask.CLEAN_UP_LIVING_ROOM,
+                    AvailableTask.CLEAN_UP_BEDROOM,
+                    AvailableTask.CLEAN_UP_BATHROOM,
+                )
+            ]
+        case AvailableTask.MULTI_TASK_8:
             return [
                 keep_only_n_scenes(task_blueprints_configs[task], nb_scenes)
                 for task in (
@@ -465,6 +476,10 @@ def get_task_blueprint_config(task: AvailableTask, nb_scenes: int) -> list[dict[
                     AvailableTask.RELAX_ON_SOFA,
                     AvailableTask.READ_BOOK_IN_BED,
                     AvailableTask.SETUP_BATH,
+                    AvailableTask.CLEAN_UP_KITCHEN,
+                    AvailableTask.CLEAN_UP_LIVING_ROOM,
+                    AvailableTask.CLEAN_UP_BEDROOM,
+                    AvailableTask.CLEAN_UP_BATHROOM,
                 )
             ]
         case _:
diff --git a/examples/benchmark/train.py b/examples/benchmark/train.py
index 9742c14..35dc680 100644
--- a/examples/benchmark/train.py
+++ b/examples/benchmark/train.py
@@ -124,7 +124,7 @@ def main(
         do_eval (bool): Evaluate the agent. !! Don't eval with a different environment in a Docker container, both rendering windows might be mixed up.
         randomize_agent_position (bool): Randomize the agent position in the environment.
     """
-    is_single_task = task != AvailableTask.MULTI_TASK
+    is_single_task = task not in {AvailableTask.MULTI_TASK_4, AvailableTask.MULTI_TASK_8}
     if is_single_task:
         model_config["policy_type"] = "CnnPolicy"
     else: