diff --git a/ai/litellm-config.yaml b/ai/litellm-config.yaml
index 15e0c02..4a61bf0 100644
--- a/ai/litellm-config.yaml
+++ b/ai/litellm-config.yaml
@@ -39,7 +39,7 @@ model_list:
       tpm: 100000
       timeout: 600  # 10 minutes for generation
       stream_timeout: 600
-      supports_system_messages: false  # vLLM handles system messages differently
+      supports_system_messages: true  # Qwen supports system messages
       stream: true  # Enable streaming by default
 
   - model_name: llama-3.1-8b