Status

Operational

Last check 15 Nov 20:00 UTC. Checks run every 5 minutes. The same machine-readable data is at /status.json; /v1/health needs no key.

Latency, median (24 h): 200 ms
Latency, p95 (24 h): 774 ms
Uptime, 30 days: 99.74%
Uptime, 90 days: 99.74%

Hourly p95 latency, last 24 hours (highest 1121 ms)

Latency by endpoint

Endpoint	Median (24 h)	p95 (24 h)	Median (last hour)	p95 (last hour)
GET /v1/recalls/{id}	317 ms	827 ms	211 ms	718 ms
GET /v1/search	420 ms	1275 ms	410 ms	1241 ms
GET /v1/barcode/{code}	100 ms	266 ms	70 ms	238 ms
GET /v1/vehicles/{make}/{model}/{year}	102 ms	277 ms	72 ms	249 ms
GET /v1/health (over the public network)	326 ms	642 ms	323 ms	612 ms
Data service health	307 ms	800 ms	205 ms	940 ms
API database	121 ms	267 ms	106 ms	278 ms

Each row is a check the API runs on itself every 5 minutes, using the same code a customer request uses (without the key checks). Only successful checks are timed.

Data sources

Source	Last successful refresh	State	Refresh time (UTC)
NHTSA	15 Nov 20:02 UTC	current	10:00 daily
CPSC	15 Nov 19:47 UTC	current	09:30 daily
FDA	15 Nov 19:52 UTC	current	09:30 and 09:35 daily

A source is "stale" when it has not refreshed within its window. A refresh that fails or comes back empty never replaces good data, and is not counted as a success.

Incidents

No incidents in the last 90 days.

An incident opens when the checks fail for 10 minutes or more and closes on its own when they pass again. History is kept 90 days.