diff --git a/vllm/model_executor/guided_decoding/xgrammar_decoding.py b/vllm/model_executor/guided_decoding/xgrammar_decoding.py
index 043d33ed96665..e93a0bd7b268c 100644
--- a/vllm/model_executor/guided_decoding/xgrammar_decoding.py
+++ b/vllm/model_executor/guided_decoding/xgrammar_decoding.py
@@ -233,6 +233,8 @@ def __call__(self, input_ids: list[int],
 
         for i, matcher in enumerate(self.matchers):
             if not matcher.is_terminated():
+                # @ubospica: ideally, fill_next_token_bitmask should be parallelized with model decoding
+                # See https://github.com/vllm-project/vllm/pull/10785/files#r1864278303
                 matcher.fill_next_token_bitmask(self.token_bitmask, i)
 
         # token_bitmask is a CPU tensor for use with accept_token and