Summarize a request log into p50/p95 latency, total cost, cost per 1k tokens, and a budget breach check. Runs entirely in your browser.
Bundled sample request log (edit in src/app.ts and rebuild to try your own):
Known limitations: linear-interpolation percentiles on a tiny sample; budget
thresholds are illustrative constants. See the project README for the full write-up and the Python CLI version.