diff --git a/vllm/v1/attention/backends/triton_attn.py b/vllm/v1/attention/backends/triton_attn.py
index ecb92bb1e4161..4c5a1a755c1a6 100644
--- a/vllm/v1/attention/backends/triton_attn.py
+++ b/vllm/v1/attention/backends/triton_attn.py
@@ -376,7 +376,7 @@ class TritonAttentionImpl(AttentionImpl):
                     query.reshape(
                         (num_tokens, num_heads * head_size)).contiguous(),
                     layer._q_scale)
-            query = query.reshape((num_tokens, num_heads, head_size))
+                query = query.reshape((num_tokens, num_heads, head_size))
 
         use_local_attn = \
             (self.use_irope and attn_metadata.local_attn_metadata is not None)