diff --git a/llama_cpp/llama_cache.py b/llama_cpp/llama_cache.py
index e059e98e1b..4d64443cc2 100644
--- a/llama_cpp/llama_cache.py
+++ b/llama_cpp/llama_cache.py
@@ -108,7 +108,7 @@ def __init__(
         self, cache_dir: str = ".cache/llama_cache", capacity_bytes: int = (2 << 30)
     ):
         super().__init__(capacity_bytes)
-        self.cache = diskcache.Cache(cache_dir)
+        self.cache = diskcache.Cache(cache_dir, size_limit=capacity_bytes)
 
     @property
     def cache_size(self):