diff --git a/src/h3_blackwell_runtime/qwen3vl_vision.py b/src/h3_blackwell_runtime/qwen3vl_vision.py
index 7fd2fcb..97cd465 100644
--- a/src/h3_blackwell_runtime/qwen3vl_vision.py
+++ b/src/h3_blackwell_runtime/qwen3vl_vision.py
@@ -103,7 +103,7 @@ def process_image(
 
     grid_h = h_bar // patch_size
     grid_w = w_bar // patch_size
-    grid_thw = torch.tensor([1, grid_h, grid_w], device=device, dtype=torch.long)
+    grid_thw = torch.tensor([[1, grid_h, grid_w]], device=device, dtype=torch.long)
 
     pixel_values = normalized
     channel = pixel_values.shape[0]