diff --git a/vllm/entrypoints/openai/logits_processors.py b/vllm/entrypoints/openai/logits_processors.py
index 7913f8720ca73..c8132811de903 100644
--- a/vllm/entrypoints/openai/logits_processors.py
+++ b/vllm/entrypoints/openai/logits_processors.py
@@ -71,7 +71,7 @@ def get_logits_processors(
 
         # Check if token_id is within the vocab size
         for token_id, bias in clamped_logit_bias.items():
-            if token_id < 0 or token_id >= tokenizer.vocab_size:
+            if token_id < 0 or token_id >= len(tokenizer):
                 raise ValueError(f"token_id {token_id} in logit_bias contains "
                                  "out-of-vocab token id")
 
@@ -81,6 +81,6 @@ def get_logits_processors(
     if allowed_token_ids is not None:
         logits_processors.append(
             _get_allowed_token_ids_logits_processor(
-                frozenset(allowed_token_ids), tokenizer.vocab_size))
+                frozenset(allowed_token_ids), len(tokenizer)))
 
     return logits_processors