Usage meter — watch your quota being spent, live

Pan and zoom the map below. Every tile it fetches is counted in your browser, in real time — requests, quota units, requests-per-second, burst — exactly the way our metering counts them. Switch mode to feel the difference between vector, raster and retina. Nothing here touches your real quota: internal traffic is unmetered.

0tile requests this session
0quota units consumed
0requests / second now
0peak burst (req in 1 s)
–avg units / map action

Requests per second — last 60 s

Each bar is one second. A map load or fast zoom fires a short burst — this is why plans have a sustained rate + burst allowance, not a bare per-second cap.

Would each plan have kept up with this session?

PlanSustainedBurstMonthly unitsYour peak burstVerdictThis session ≈
Verdict compares your peak 1-second burst and 10-second average against each plan's token bucket. "This session" projects how many visitors behaving like you fit in the monthly quota.

How the numbers work

Every tile request that reaches our servers counts, weighted by what it costs us to serve: vector 1 · raster 3 · retina 5 units (rendering raster is ~1000× more expensive than streaming vector bytes — the deep guide explains why). A typical map load fetches 6–16 tiles; each pan wave adds a few more. Rate limits are per account (keys share them) as a token bucket: you can burst well above the sustained rate for a moment — exactly what a map load needs. Over quota: Free pauses until next month, paid tiers get +10% grace. Your own CDN's cache hits never reach us and are free.