AI provider latency and uptime API — data only we have.
How fast and how reliable each AI provider is right now — latency and uptime we measure ourselves, continuously, into the append-only signed archive.
The call
Keyless over MCP, or one REST call with a free key. Every datapoint is provenance-stamped and citeable.
# REST (free key)
curl -H "X-API-Key: $KEY" "https://dynamicfeed.ai/benchmarks"
# MCP (keyless) — point any client at https://dynamicfeed.ai/mcp, then call:
ai_provider_latency()Sample response
{ "provider": "example-ai", "p50_ms": 420, "p95_ms": 980,
"uptime_24h_pct": 99.7, "observed_at": "2026-06-03T13:00:00Z" }Why live data
No model knows how its provider is performing this minute — and this archive can't be reconstructed after the fact. It is the one proprietary, un-backfillable feed in Dynamic Feed.
Use it for
- Model-router and failover agents
- SLA and reliability dashboards
- Provider selection and cost-vs-latency tradeoffs
FAQ
Where does this come from?
We measure it ourselves on a fixed schedule — proprietary, not a third-party feed.
Is there history?
The series is retained in the append-only signed archive; historical export is available for research / eval.
Keyless?
MCP is keyless; REST uses a free key.