~10-15 min on A10G ($0.30-0.50). pi0 LoRA is 3-5x slower due to the bigger backbone — budget accordingly. hf_cache = modal.Volume.from_name("pi0-hf-cache", create_if_missing=True) onnx_output = ...

You should see your app running at http://localhost:3000. Now you can modify the source code in my_app_name/my_app_name.py. Reflex has fast refreshes so you can see ...