diff --git a/examples/server/server.cpp b/examples/server/server.cpp
index b0f0486b7..37b5b78d3 100644
--- a/examples/server/server.cpp
+++ b/examples/server/server.cpp
@@ -52,7 +52,7 @@ struct llama_server_context
     n_consumed = 0;
   }
 
-  bool loadModel(gpt_params params_)
+  bool loadModel(const gpt_params &params_)
   {
     params = params_;
     ctx = llama_init_from_gpt_params(params);