I'm loving the tit for tat cost charts these guys are doing. Just a few days ago it looked like OpenAI ruled the cost pareto frontier. Not even a week later and Anthropic is taking the charts again. See you guys same time next week?
Anyway I think if you have a single stream of a cheap model, like GPT 6 Luna, I don't think you can currently exhaust it in a week on a $200 plan. I mean it only puts out so many tokens per second.