I'd wait to see if it tapers back down before doing anything. Spiked usage is a pretty common thing for LLM apps.

If it persists or you think it will, reach out to me at parham@bearly.ai and we'd be interested to sponsoring it or helping how we can :)