Show HN: Does per-step reasoning effort save money in Claude Code?

2 comments

• ethan1998 3 days ago

  If you change reasoning during a session, LLM needs to recompute KV cache. Maybe I get it wrong, but I don't think it can save money.

• ifoster41901 3 days ago

  [flagged]