diff --git a/llama_cpp/llama_chat_format.py b/llama_cpp/llama_chat_format.py
index 17575c700..18caa1a2e 100644
--- a/llama_cpp/llama_chat_format.py
+++ b/llama_cpp/llama_chat_format.py
@@ -594,7 +594,8 @@ def chat_completion_handler(
             tool_choice=tool_choice,
         )
         prompt = llama.tokenize(
-            result.prompt.encode("utf-8"),
+            vocab=llama.llama_model_get_vocab(model),
+            text=result.prompt.encode("utf-8"),
             add_bos=not result.added_special,
             special=True,
         )