load model when init

app.py

@@ -61,6 +61,7 @@ _LOADED_MODEL_NAME = None
 
 
 
+@spaces.GPU
 def get_model(model_name: Optional[str] = None):
     """Load and cache the Step3-VL-10B model and processor."""
     global _MODEL, _PROCESSOR, _LOADED_MODEL_NAME
@@ -100,6 +101,13 @@ def get_model(model_name: Optional[str] = None):
     return _MODEL, _PROCESSOR
 
 
+@spaces.GPU
+def warmup_model(model_name: Optional[str] = None):
+    """Warmup the model once at app startup."""
+    get_model(model_name)
+    return True
+
+
 def resolve_generation_tokens(processor):
     """Resolve eos/pad token ids and stop strings for safer stopping."""
     tokenizer = getattr(processor, "tokenizer", processor)
@@ -633,6 +641,7 @@ h3 {
 
 # Gradio Interface
 with gr.Blocks(title="Step3-VL-10B", css=custom_css, theme=gr.themes.Soft()) as demo:
+    model_ready = gr.State(False)
     # Header
     gr.HTML("""
         <div class="app-header">
@@ -745,6 +754,12 @@ with gr.Blocks(title="Step3-VL-10B", css=custom_css, theme=gr.themes.Soft()) as
         outputs=[chatbot, user_text, image_file]
     )
 
+    demo.load(
+        fn=warmup_model,
+        inputs=None,
+        outputs=model_ready
+    )
+
 if __name__ == "__main__":
     import argparse
     parser = argparse.ArgumentParser()