Paste your prompt, pick a compression level, and get an optimized version that costs less to run — with token counts and savings calculated across 25+ models instantly.