Run side by side comparisons of two models on Amazon Bedrock using Deepseek with practical checks for Llama Claude and OpenAI models