API Latency Explorer
Fire requests at an endpoint and see the percentile spread, not just the average.
IntermediateWeb appData viz8–16hAI-buildable 5/5no setupworth shipping
A developer tool that makes a point: averages lie about latency. Send a configurable burst of requests to a URL, then plot the results as a histogram with p50, p95 and p99 marked. Watching the long tail appear is the whole lesson, and the visualisation work is genuinely interesting.
UI
Logic
Useful
Hosting · Free forever
Deploys free on GitHub Pages, Cloudflare Pages or Vercel Hobby.
Deploy it and share a link.
Core features · 7
- Enter a URL and a request count
- Requests fire with a configurable concurrency limit
- A histogram renders the latency distribution
- p50, p95 and p99 markers overlay the histogram
- A live counter shows progress while the burst runs
- Failed requests are counted separately rather than skewing the timings
- The run can be cancelled part-way
The interesting part
- Percentile calculation that is correct for small sample sizes
- A concurrency limiter that doesn't just fire everything at once
- Cancelling in-flight requests cleanly with AbortController
- CORS will block most third-party URLs — the UI has to explain that clearly
Once it works
- Compare two endpoints side by side
- Persist past runs so results can be compared over time
- Add a warm-up phase excluded from the stats
Suggested stack
reacttypescripttailwindrecharts