diff --git a/ansible/roles/llm-inference-multimodel/defaults/main.yml b/ansible/roles/llm-inference-multimodel/defaults/main.yml
index 74ef9e8..377659c 100644
--- a/ansible/roles/llm-inference-multimodel/defaults/main.yml
+++ b/ansible/roles/llm-inference-multimodel/defaults/main.yml
@@ -71,7 +71,7 @@ llm_toolcall_model_id: mistral-small-24b-instruct-2501
 # (either free VRAM by stopping llama-server-toolcall for the shadow window,
 # or reduce ctx-size). llm_qwen_service_enabled defaults to false so a normal
 # playbook run will template the unit but NOT start it.
-llm_qwen_service_enabled: true
+llm_qwen_service_enabled: false
 llm_qwen_port: 8002
 llm_qwen_model_path: "{{ llm_models_dir }}/Qwen2.5-14B-Instruct-Q5_K_M.gguf"
 llm_qwen_model_url: "https://huggingface.co/bartowski/Qwen2.5-14B-Instruct-GGUF/resolve/main/Qwen2.5-14B-Instruct-Q5_K_M.gguf"