Cache misses are the one reason I stopped using OpenCode in favor of Pi. OpenCode mutates the system prompt every turn, which is completely unacceptable and betrays such a basic misunderstanding of LLMs that I can't in good conscience trust the rest of the product regardless of how technically impressive it is.

I've had the same experience. Tried it and use in 2 hours more token than in a whole week. Normally i got 80 - 90 % cache hit rate.