diff --git a/vllm/v1/engine/detokenizer.py b/vllm/v1/engine/detokenizer.py
index 04ad51aae0a8..0ccbe6549349 100644
--- a/vllm/v1/engine/detokenizer.py
+++ b/vllm/v1/engine/detokenizer.py
@@ -234,7 +234,7 @@ class FastIncrementalDetokenizer(BaseIncrementalDetokenizer):
         try:
             token = self.stream.step(self.tokenizer, next_token_id)
         except Exception as e:
-            if str(e) != INVALID_PREFIX_ERR_MSG:
+            if not str(e).startswith(INVALID_PREFIX_ERR_MSG):
                 raise e
             # Recover from edge case where tokenizer can produce non-monotonic,
             # invalid UTF-8 output, which breaks the internal state of
@@ -243,7 +243,8 @@ class FastIncrementalDetokenizer(BaseIncrementalDetokenizer):
             logger.warning(
                 "Encountered invalid prefix detokenization error"
                 " for request %s, resetting decode stream.", self.request_id)
-            self.stream = DecodeStream(self.skip_special_tokens)
+            self.stream = DecodeStream(
+                skip_special_tokens=self.skip_special_tokens)
             token = self.stream.step(self.tokenizer, next_token_id)
         return token