HomeService Level Agreement — MCLY Technology

SERVICE

Service Level Agreement — MCLY Technology

Availability targets, latency objectives and the credit mechanism that applies when we miss them.

Last updated September 2026

1. Covered services

This agreement covers the MCLY gateway API (the /v1 endpoints) and reserved compute instances. It does not cover the storefront website, which is informational, or outages caused by an upstream provider that affect that provider across all of its customers and regions simultaneously.

2. Availability target

ServiceMonthly targetMeasurement
Gateway API (chat, embeddings)99.9%per calendar month
Model listing & account endpoints99.5%per calendar month
Reserved compute instances99.5%per calendar month

Availability is calculated as the share of minutes in the month during which the service returned a valid response, excluding scheduled maintenance announced at least 72 hours in advance and windows of up to 30 minutes used for emergency security patching.

3. Latency objectives

Routing overhead added by the gateway — the time between receiving your request and dispatching it upstream, plus the return path — is targeted at a median under 400 ms. Time spent generating tokens upstream is outside our control and is reported separately on each response as latency_ms.

4. Support response targets

SeverityDefinitionFirst response
CriticalProduction traffic failing or fully unavailable1 hour
HighPartial degradation, sustained elevated errors or latency4 hours
NormalIntegration questions, quota changes, billing1 business day

Targets apply during business hours in US Mountain Time; critical issues are monitored around the clock.

5. Service credits

Monthly availabilityCredit
Below 99.9% but at or above 99.0%10% of monthly fees
Below 99.0% but at or above 98.0%25% of monthly fees
Below 98.0%50% of monthly fees

Credits are issued on request within 30 days of the affected month, applied to future invoices, and are not convertible to cash. Credits are the exclusive remedy for missed availability targets.

6. Exclusions

  • Your own application, network or credential problems, including exhausted quotas and revoked keys;
  • models or providers marked in our catalogue as experimental or preview, which are offered without an availability target;
  • suspension of an account for abuse, breach of the terms of service, or non-payment;
  • force majeure, including large-scale internet outages and actions of upstream providers that affect the wider market.

7. Reporting

Report an incident to [email protected] with the request IDs or order number concerned, the model name and the UTC window of the impact. Reports without identifying details take longer to substantiate.

Start routing in minutes

One endpoint. Every frontier model.

Create an account, pick a plan, and point your existing OpenAI client at our base URL. No SDK rewrite, no lock-in.