diff --git a/modelscope/cli/llamafile.py b/modelscope/cli/llamafile.py
index fab7a193..39d1346f 100644
--- a/modelscope/cli/llamafile.py
+++ b/modelscope/cli/llamafile.py
@@ -145,6 +145,8 @@ class LlamafileCMD(CLICommand):
             # if torch is not available, we will just assume gpu cannot be used
             pass
         if has_gpu:
+            print(
+                'GPU detected, launching model with llamafile GPU option >>>')
             execute_cmd = f'{execute_cmd} -ngl 999'
         os.system(execute_cmd)