Service Level Agreement
Last updated Jan 14, 2026 · US Eastern Time (Washington, D.C.)
1. Scope
This SLA covers the production API endpoints /api/v1/search and
/api/v1/url. It does not cover the marketing website, documentation, or beta
features explicitly labeled as previews or placeholders.
2. Availability target
| Plan | Monthly availability target | Support response |
|---|---|---|
| Standard | 99.0% | 2 business days |
| Starter | 99.5% | 1 business day |
| Pro | 99.9% | 8 business hours |
| Ultimate | 99.95% | 4 business hours |
Availability is measured per calendar month in US Eastern Time (America/New_York). Business hours are 09:00-18:00 US Eastern Time, Monday to Friday, excluding US federal holidays.
3. What counts as downtime
A minute is counted as downtime when requests that are correctly authenticated and correctly formed fail with a 5xx status, for reasons within our control. The following are excluded:
- Scheduled maintenance announced at least 48 hours in advance.
- Rate limiting (HTTP 429) triggered by your own traffic.
- Failures caused by the target website being slow, blocked or unavailable.
- Force majeure events and upstream network outages outside our infrastructure.
4. Latency expectations
Each request carries its own latency budget through the d parameter. We honor the
budget you set: if results can't be produced within it, the call returns an error rather than
running long. Credits are not consumed for failed calls.
5. Remedies
| Measured monthly availability | Credit remedy |
|---|---|
| Below target, at or above 98% | 10% of that month's consumed credits returned |
| Below 98%, at or above 95% | 25% of that month's consumed credits returned |
| Below 95% | 50% of that month's consumed credits returned |
Claims must be submitted to support@serpforai.dev within 30 days of the incident and include request identifiers or timestamps.
6. Current status
The production backend is not publicly deployed yet. Until launch this SLA describes the commitments that will apply at general availability; no availability claims can be made against the pre-launch environment.