The model container can take a while to spin up- it's downloading a Qwen2 from Huggingface and saving it into an emulated AWS data connection.