import gradio as gr def calculate_tco(model_choice, vm_rental_choice): VM_cost_per_hour=3.6730 tokens_per_request = 64 if model_choice == "Llama-2-7B": tokens_per_second=694.38 elif model_choice == "Llama-2-13B": tokens_per_second=1000 elif model_choice == "Llama-2-70B": tokens_per_second=10000 if vm_rental_option == "1 month" or "3 months" or "6 months": reduction = 0 elif vm_rental_option == "1 year": reduction = 0.34 elif vm_rental_option == "3 years": reduction = 0.62 homemade_cost_per_request = VM_cost_per_hour * (1 - reduction) * tokens_per_request / (tokens_per_second * 3600) #* 0.8 * 0.5 saas_cost_per_request = 0.018 * tokens_per_request output = f"Cost/request with a plan for {vm_rental_choice} using:\n" output += f"- SaaS solution from OpenAI: ${saas_cost_per_request:.5f}\n" output += f"- Home-made solution with the model {model_choice}: ${homemade_cost_per_request:.5f}" return output models = ["Llama-2-7B", "Llama-2-13B", "Llama-2-70B"] vm_rental_option = ["1 month", "3 months", "6 months", "1 year", "3 years"] iface = gr.Interface( fn=calculate_tco, inputs=[gr.inputs.Dropdown(models, label="Select AI Model"), gr.inputs.Dropdown(vm_rental_option, label="Select VM Rental Duration")], outputs=gr.outputs.Textbox() ) iface.launch()