The Real-time Updates API is currently released in open beta, it is considered stable and is capable of supporting hudreads of thousands of concurrent connections.
Since this API is meant to be used for delivering content from DatoCMS to apps, websites and other digital products consumed by end-users, there are no limits enforced on the number of concurrent connections, although a fair use policy of a maximum of 1k open connections per project applies.
If you're planning to have more than 1k concurrent connections open for long periods of time, please do contact us so we can discuss together about your needs and see how we can help you scaling.
Even though the Real-time Updates API is not billed per se, it uses the Content Delivery API to work, so it contributes to the number of API calls to it.
Indicatively, as many requests will be made to the Content Delivery API as there are update events received on a channel, however, the count of requests is largely independent of the number of connected clients, as long as they are all connected to exactly the same GraphQL query, with the same API token, on the same environment and draft/published version.
Requests made to CDA/minute = Rate of updates/minute * N
N is proportional to the overall number of open connections (the bigger the number of connections, the higher it will be), but is capped to a maximum of 15.
Let's illustrate the count by using an example:
the home page of a site makes a connection to the Real-time Updates API;
the GraphQL query and API token are hard-coded in the code of the webpage, so they're the same for all;
the content of the page is modified by editors approx. once every 30 seconds;
a total of 500 users is connected at the same time to the homepage in question.
The number of API calls made to the Content Delivery API in 10 minutes will be between 20 and 300.