What would your bill look like with the noise compressed out?
Upload a usage export from OpenAI, Anthropic, GitHub Copilot, or Cursor — or just type your monthly spend — and see estimated savings in tokens and dollars, before and after.
Runs entirely in your browser. Your usage data is never uploaded anywhere.
Enter your monthly spend as a number, e.g. 240.
Drop your usage file here or click to browse — CSV or JSON
OpenAI usage/cost exports · Anthropic Console usage · Claude Admin API usage report (JSON) · Copilot premium-request report · Cursor usage · any CSV with token or cost columns
Couldn’t find token or cost data in that file. Try the quick estimate instead.
Estimated savings
Conservative to optimistic: $0 – $0 per month · per year at the typical rate
Like these numbers? Early access makes them real.
Enter a valid email so we can reach you.
You’re on the list. Check your email for a note from us.
Tokens per month, before and after
Typical estimate; hover a bar for the range.
Monthly cost, before and after
Typical estimate; hover a bar for the range.
How this estimate works
Compression applies to the input side of your traffic — tool output, file reads, logs, search results — which is where coding-agent bills concentrate. We assume ~85% of input traffic is compressible and apply the measured 60–90% reduction to that slice (60% conservative, 75% typical, 90% optimistic). Output tokens are never touched.
That’s why the whole-bill number here is lower than the headline 60–90%: the headline is what happens to compressible tool-call traffic; this page shows the honest blended effect on everything you pay for.
Want these numbers for real, not estimated?
Join the waitlist and we’ll measure your actual traffic during early access — same agent, same answers, smaller bill.
Enter a valid email so we can reach you.
No spam, just an early-access invite when your spot opens.
You’re on the list. Check your email for a note from us.