diff --git a/examples/foundational/12-describe-video.py b/examples/foundational/12-describe-video.py
index 8e20d6533..62e116020 100644
--- a/examples/foundational/12-describe-video.py
+++ b/examples/foundational/12-describe-video.py
@@ -61,6 +61,7 @@ async def main(room_url: str, token):
 
         vision_aggregator = VisionImageFrameAggregator()
 
+        # If you run into weird description, try with use_cpu=True
         moondream = MoondreamService()
 
         tts = ElevenLabsTTSService(
diff --git a/src/dailyai/services/moondream_ai_service.py b/src/dailyai/services/moondream_ai_service.py
index 07ff9e534..2b9835404 100644
--- a/src/dailyai/services/moondream_ai_service.py
+++ b/src/dailyai/services/moondream_ai_service.py
@@ -25,11 +25,11 @@ def __init__(
         self,
         model_id="vikhyatk/moondream2",
         revision="2024-04-02",
-        device=None
+        use_cpu=False
     ):
         super().__init__()
 
-        if not device:
+        if not use_cpu:
             device, dtype = detect_device()
         else:
             device = torch.device("cpu")