Skip to content

SLAs

This page describes the service-level commitments for oneapi.finance. They are practical commitments, not contractual SLAs with credit clauses. Custom enterprise contracts can include stricter terms; talk to us.

Availability target

TierUptime targetWindow measuredWhat “down” means
Free99.0%Calendar month, UTCThe gateway returns 5xx for >5% of requests for 5+ consecutive minutes.
Indie / Pro99.5%Calendar month, UTCSame as above.
Business99.9%Calendar month, UTCSame as above.
CustomPer contractPer contractPer contract.

A 99.5% target permits ~3.6 hours of downtime per month. The bulk of any downtime in practice is upstream-source related, not gateway related; see data sources for failover behavior.

Latency target

Endpointp50p95p99
/v1/quote (cache hit)30 ms80 ms200 ms
/v1/quote (cache miss, single upstream)250 ms800 ms2,500 ms
/v1/time_series (cache hit)50 ms150 ms400 ms
/v1/time_series (cache miss, daily)400 ms1,500 ms4,000 ms
/v1/statistics (cache hit)30 ms80 ms200 ms
/v1/symbol_search80 ms200 ms500 ms

Numbers are gateway-internal latency, EU and US east regions. Add ~100 ms for APAC clients reaching our European POP. We do not currently advertise edge deployment in Asia.

Freshness target

DataTarget latency from source publicationNotes
US equity quotes≤30 minutes15-min delay at the source plus our cache window.
Non-US equity quotes≤30 minutesSame.
Daily OHLCV (yesterday’s close)Available by 04:00 UTC the next dayWorker schedule.
Statistics (price-derived)Same as the underlying quoteRecomputed in place.
Statistics (filing-derived)≤4 hours from EDGAR acceptanceWorker schedule.
Profile≤7 days from public updateWeekly refresh.
Dividends/splits≤24 hours from announcementDaily worker.
FX≤24 hours from publicationDaily fixings.
Crypto≤2 minutesCoinGecko polling.

What we do when we miss

If we breach an availability target in a calendar month, we will:

  1. Post a public post-mortem at status.oneapi.finance within seven days.
  2. Email Indie+ customers with a summary and root cause.
  3. Offer a credit for the affected customer’s next month, on request.

We do not auto-issue credits. If you were affected and want one, email support@oneapi.finance.

What is not covered

  • Upstream-source outages. When all three of our quote sources are down simultaneously, we return 502 upstream_failure. Those minutes count against availability for transparency, but they are not always fixable on our end. We mitigate by adding sources, not by lying about uptime.
  • Network paths outside our control. ISP outages, DNS misconfigurations on the client side, BGP incidents.
  • Customer-side rate limits. A 429 rate_limit is a feature, not an outage.
  • Free-tier abuse mitigation. If your IP is throttled because you are running 10,000 requests per minute through 50 free-tier keys, that is intended behavior, not an SLA breach.

What we do not promise

  • Real-time prices. See delayed data.
  • 100% data coverage. Some symbols, especially small-cap non-US listings, will simply be missing. We add coverage in response to requests; file an issue if a symbol you care about is unavailable.
  • Backwards compatibility forever. We promise 24-month deprecation windows for major versions and a published changelog for every change. We do not promise the response shape will never evolve.

Custom contracts

For organizations that need stronger SLAs, dedicated capacity, or per-client support: we are happy to discuss a custom contract. Email contracts@oneapi.finance with traffic shape, peak QPS, and the endpoints you depend on.

See also