diff --git a/src/lerobot/envs/vlabench.py b/src/lerobot/envs/vlabench.py
index ea72b77cb..1a9ed024d 100644
--- a/src/lerobot/envs/vlabench.py
+++ b/src/lerobot/envs/vlabench.py
@@ -344,19 +344,6 @@ class VLABenchEnv(gym.Env):
             dtype=np.float64,
         )
 
-    @staticmethod
-    def _normalize_gripper_action(gripper: float) -> float:
-        """Normalize the scalar gripper command to VLABench's [0, 1] convention.
-
-        The native VLABench collector uses 0=open and 1=closed. Older LeRobot
-        integrations often assumed a symmetric [-1, 1] gripper channel, so we
-        preserve backward compatibility by remapping negative values from
-        [-1, 1] -> [0, 1] before clipping.
-        """
-        if gripper < 0.0:
-            gripper = 0.5 * (float(np.clip(gripper, -1.0, 1.0)) + 1.0)
-        return float(np.clip(gripper, 0.0, 1.0))
-
     def _build_ctrl_from_action(self, action: np.ndarray, ctrl_dim: int) -> np.ndarray:
         """Convert a 7D EEF action into the `ctrl_dim`-sized joint command vector.
 
@@ -378,7 +365,7 @@ class VLABenchEnv(gym.Env):
 
         pos = np.asarray(action[:3], dtype=np.float64)
         rx, ry, rz = float(action[3]), float(action[4]), float(action[5])
-        gripper = self._normalize_gripper_action(float(action[6]))
+        gripper = float(np.clip(action[6], 0.0, 1.0))
         quat = self._euler_xyz_to_quat_wxyz(rx, ry, rz)
 
         assert self._env is not None
@@ -401,8 +388,7 @@ class VLABenchEnv(gym.Env):
 
         # Gripper: action scalar in [0, 1] (0=open, 1=closed). Map linearly to
         # finger qpos in [CLOSED, OPEN]. Franka has 2 mirrored fingers.
-        g = gripper
-        finger_qpos = self._FRANKA_FINGER_OPEN + g * (self._FRANKA_FINGER_CLOSED - self._FRANKA_FINGER_OPEN)
+        finger_qpos = self._FRANKA_FINGER_OPEN + gripper * (self._FRANKA_FINGER_CLOSED - self._FRANKA_FINGER_OPEN)
 
         ctrl = np.zeros(ctrl_dim, dtype=np.float64)
         ctrl[:n_dof] = arm_qpos
diff --git a/tests/envs/test_vlabench_env.py b/tests/envs/test_vlabench_env.py
deleted file mode 100644
index 97456cc57..000000000
--- a/tests/envs/test_vlabench_env.py
+++ /dev/null
@@ -1,40 +0,0 @@
-#!/usr/bin/env python
-
-# Copyright 2025 The HuggingFace Inc. team. 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.
-
-import numpy as np
-
-from lerobot.envs.vlabench import ACTION_HIGH, ACTION_LOW, VLABenchEnv
-
-
-def test_vlabench_action_space_uses_zero_to_one_gripper_channel():
-    env = VLABenchEnv()
-
-    np.testing.assert_array_equal(env.action_space.low, ACTION_LOW)
-    np.testing.assert_array_equal(env.action_space.high, ACTION_HIGH)
-    assert env.action_space.shape == (7,)
-    assert env.action_space.low[-1] == 0.0
-    assert env.action_space.high[-1] == 1.0
-    np.testing.assert_array_equal(env.action_space.low[:6], np.full(6, -1.0, dtype=np.float32))
-    np.testing.assert_array_equal(env.action_space.high[:6], np.full(6, 1.0, dtype=np.float32))
-
-
-def test_vlabench_gripper_action_normalization_keeps_backward_compatibility():
-    assert VLABenchEnv._normalize_gripper_action(-1.0) == 0.0
-    assert VLABenchEnv._normalize_gripper_action(-0.5) == 0.25
-    assert VLABenchEnv._normalize_gripper_action(0.0) == 0.0
-    assert VLABenchEnv._normalize_gripper_action(0.5) == 0.5
-    assert VLABenchEnv._normalize_gripper_action(1.0) == 1.0
-    assert VLABenchEnv._normalize_gripper_action(1.5) == 1.0