Price and mitigations

[Lesaje]

Hi, I started test simulation using base_the_ville_isabella_maria_klaus, and for 40 seconds of simulation it spent ~2.6$. Are this expenditure rates sound right? If I'm correct, it's somewhere around 2.6/0.0015$/1K tokens = ~1.7 million tokens for 3 persons and 40 seconds.

If so, what's the progress on optimization and/or switching to other LLM, for example, Alpaca?

Thank you all for responses/thoughts.