Status
No status badge nobody updates. These numbers come from the service itself.
Freshness
New articles land once an hour. The last batch landed at
2026-09-07 08:57Z and yielded
13,820 articles from 21,169 crawled pages.
Full history of every batch, with timing and country breakdown, is on snapshots. That page is the honest status board: if ingest stops, it stops visibly.
Check it yourself
curl "https://freenewsapi.ai/health"{"status": "ok", "documents": 21690432}Measured performance
| Metric | Value |
|---|---|
| Sustained throughput | 1,247 req/s |
| p99 latency under load | 107 ms |
| Errors over 40,000 requests | 0 |
| Search cluster | 3 nodes, every index replicated to all three |
| Per-IP limit | 20 req/s, burst 40 |
What can break, and how you would know
Three things can fail independently, and they fail in ways that look different:
| Failure | Symptom | Where you see it |
|---|---|---|
| Search cluster unreachable | 503 on every request; /health reports
degraded |
/health, immediately |
| Ingest stalled | Search works normally, but no new articles arrive and the newest
crawled_at stops moving |
/snapshots — the gap is visible in the table |
| Upstream archive delayed | Ingest is healthy but has nothing to fetch; the same symptom as a stall, with a different cause | /snapshots, same place |
The distinction matters for an agent: a 503 means retry later, whereas stale data means the answers are still valid but the window has stopped advancing. Only the first is worth backing off for.
No SLA
There is no uptime guarantee, because there is no contract and no charge. What there is instead: a published pipeline, a public health endpoint and a visible ingest log, so you can measure reliability rather than be promised it.
