diff --git a/llama.cpp/compose.yml b/llama.cpp/compose.yml
index 1cf3342..0549b82 100644
--- a/llama.cpp/compose.yml
+++ b/llama.cpp/compose.yml
@@ -5,7 +5,7 @@ services:
     ports:
       - 8878:8080
     volumes:
-      - /home/andy/.lmstudio/models/lmstudio-community:/root/.cache/llama.cpp/
+      - /home/andy/.lmstudio/models/lmstudio-community:/models
       # Optionally mount a local .gguf file directly:
       # - /absolute/path/to/your-model.gguf:/model.gguf:ro
     devices:
@@ -33,6 +33,7 @@ services:
       
     restart: unless-stopped
     command: >
+      --models-dir /models
       --n-gpu-layers all
       --parallel 2
       --flash-attn on