I'm thinking if this is a more fair comparison among other models, like ChatGPT, Claude and Deepseek