Service-Level Agreement

Service-level agreement (SLA) for the ModelFront API and console

We offer two main services, the risk prediction API and the console. We offer the following default service-level agreement.

Security, privacy and data confidentiality

Our first priority is security - protecting your data and your clients' or users' data. You can read more on how we approach security, privacy and data confidentiality at modelfront.com/security

Terms of service

You can read our terms of service at modelfront.com/terms.

Uptime

ModelFront guarantees a 99.9% monthly uptime percentage threshold for both the console and API to clients and partners with a contract, subscription or invoicing agreement in-place.

If our uptime falls below this threshold, assuming no extraordinary events and no abuse nor use of the services that goes against the ModelFront instructions, we will grant service credit to the affected client for future use, taking into account the severity of the downtime and the usage of the client.

Third-party integrations

ModelFront cannot guarantee the uptime of our infrastructure providers, Google Cloud and Cloudflare, nor the availability of their core resources, although their unavailability is considered an extraordinary event.

ModelFront cannot guarantee the uptime or latency of integrations with third-party translation APIs, which can go down or change at any time.

Failed requests

ModelFront makes a best effort to safely and stabley handle all characters, and to error quickly and clearly on invalid input. Failed requests are not counted for billing purposes.

Latency

ModelFront makes a best effort to provide API response time within the same order of magnitude as major machine translation APIs.

Accuracy

Because definitions of translation quality and risk across all language pairs, content types and use cases are subjective and infinitely variable, ModelFront cannot guarantee accuracy of the translation risk predictions. You should not use ModelFront in ways that result in translations that are dangerous or illegal.


Incident history

11 March 2021

A legacy Kubernetes configuration was accidentally deployed to production. ModelFront API clients received slow responses or 503 error codes. ModelFront console users experienced slow evaluations.

14 December 2021

Widespread outages hit multiple Google services, including Google Translate. ModelFront was not affected.

March 2020

The COVID-19 pandemic disrupts the operations of many global businesses. ModelFront was not affected.


Updates

We continually update our service-level agreement as we grow and as technology, laws and concerns change. We last updated this service-level agreement in May 2020.