AccelByte Service Level Agreement

Last updated on July 9, 2026

This AccelByte Service Level Agreement (“SLA”) is a policy governing the use of AccelByte Services, and applies separately to each account using AccelByte Services.

Service Commitment​

Accelbyte will use reasonable efforts to make AccelByte Services available with a Monthly Uptime Percentage, as defined below, during any monthly billing cycle (the “Service Commitment”):

Covered ServiceMonthly Uptime Percentage
AGS>=99.9%
AMS>=99.9%
Extend>=99.9%

Definitions​

The following definitions apply to the SLA:

  • Core Services means the following AccelByte services: Foundations, Online and Multiplayer, as documented in the AccelByte API reference; DSHub notifications; lobby notifications; and such other services as may be added by mutual written agreement.
  • “Request” means a customer-initiated API call to a Core Service that is documented in the AccelByte API reference. For the avoidance of doubt, a Request does not include calls to internal telemetry, observability, or metrics endpoints, administrative console operations, or calls to services designated as Beta, Legacy, or Experimental in the AccelByte API reference.
  • “Error” means any Request that meets any of the following criteria:
    • (i) Returns any HTTP 5xx status code, or a functionally equivalent server-side error response in any other protocol used by the applicable Core Service (including, without limitation, gRPC status codes INTERNAL, UNAVAILABLE, UNKNOWN, and DATA_LOSS, or TCP connection failure, assuming use of the AccelByte SDK or proper implementation of websocket keepalive logic);
    • (ii) Fails due to a connection timeout, connection refusal, or other transport-level failure. For the avoidance of doubt, a failure described in clauses (ii) above shall not constitute an Error to the extent that it is caused by network conditions or infrastructure outside of AccelByte's network boundary, including Customer's internet service provider, Customer's local network, or public internet routing beyond AccelByte's last point of network egress.
  • “Unavailable” and “Unavailability” means the percentage of Requests processed by the applicable Covered Service during each 5-minute interval that constitute Errors. If no Requests are received by a Core Service during a 5-minute interval, and that Core Service reported Errors in either of the two immediately preceding 5-minute intervals, Unavailability for that interval is assumed to be 100%. Otherwise, Unavailability for that interval is assumed to be 0%.
  • “Monthly Uptime Percentage” is calculated by subtracting from 100% the average of the Unavailability for all 5-minute intervals in a monthly billing cycle. Notwithstanding the foregoing, any single continuous period of Unavailability of 100% for a Core Service lasting 30 minutes or more shall be deemed a "Critical Outage." Each Critical Outage shall independently trigger contractual remedies regardless of the overall Monthly Uptime Percentage for that month.
  • “Incident Window” defines the operational boundaries of tracking uptime:
    • Downtime shall begin when the issue first occurs, or when AccelByte becomes aware (or should reasonably have become aware) of the issue, whichever occurs first.
    • Downtime shall end only when the affected Services are fully restored and operating normally for at least ten (10) consecutive minutes.
  • “Monitoring and Source of Truth” governs data conflict tracking:
    • Downtime shall be measured using AccelByte's API integrated into the Customer's application alongside the Customer's independent monitoring systems.
    • In the event of a discrepancy, the measurement most favorable to the Customer shall apply unless AccelByte demonstrates manifest error.

SLA Exclusions​

The Service Commitment, respectively, does not apply to any unavailability, suspension or termination of Covered Service, or any other Covered Service performance issues:

  1. caused by factors outside of our reasonable control, including any force majeure event or Internet access or related problems beyond the demarcation point of the Covered Service;
  2. caused by failures, outages, or performance degradations of third-party services (e.g., cloud hosting, DNS services) not managed by AccelByte;
  3. caused by failures, outages, or performance degradations arising in non-GA (general availability) features or services marked as beta, experimental, or preview;
  4. that result from any of your actions or inactions;
  5. that result from your equipment, software or other technology;
  6. that result from your failure to adhere to the guidelines and best practices described in the documentation for the Covered Service;
  7. that result from failure to use the Covered Service within the bounds of the AccelByte’s published policies with respect to volume, rate limit, restriction, or from your bypassing of any features of the AccelByte SDK intended to manage interactions with the Covered Services (e.g., rate limitations or exponential backoff and retry behaviors);
  8. that result from application design issues that were:
    1. identified as risks with specific corrective actions recommended during an AccelByte Architecture Review, but these recommendations were not subsequently implemented by the customer; or
    2. related to areas typically assessed in an AccelByte Architecture Review, but the customer declined participation in such a review when it was offered.
  9. during a scheduled maintenance window;
  10. that result from a DDoS attack;
  11. that are caused by your failure to run a supported version of the environment, in accordance with the AGS Version Support Lifecycle Policy, for instance, by rejecting a release/update/hotfix;
  12. for services that are not included in AccelByte API reference;
  13. for services that are included in the AccelByte API reference that are explicitly designated or marked as “Legacy”.

The Monthly Uptime Percentage is calculated based on the definitions and standard exclusions outlined in this SLA. Should unforeseen circumstances beyond these standard exclusions impact availability, we will assess these events reasonably and in good faith to ensure the uptime calculation accurately reflects the platform's performance.

Last updated on
On this page