What persistent realtime connections cost per month.
Realtime cost splits between how many clients are connected and how much they talk. Know which one dominates before picking a plan. Compare always-on compute on the serverless cost calculator.
The WebSocket / Realtime Cost Calculator estimates your monthly realtime bill from two separate charges that most managed WebSocket services bill on. The first is your peak concurrent connections: the calculator divides that peak by 1,000 and multiplies by your price per 1,000 peak connections. The second is throughput: it multiplies your monthly message volume (in millions) by your price per million messages. Adding the two gives the total. The main drivers are therefore how many clients are connected at your busiest moment and how chatty each connection is over the month.
The key trade-off is that connections and messages scale independently, so the cheaper lever depends on your app. A dashboard with many idle sockets is dominated by the peak connection charge, while a high-frequency feed is dominated by message cost. Because connection pricing keys off the single highest peak, a brief traffic spike can set the rate for the whole period. Try trimming update frequency, batching messages, and closing idle sockets, then re-run the numbers to see which change actually moves your bill.
Managed services (Pusher, Ably, PubNub) charge on peak concurrent connections and messages sent. Both meters matter: a chat app is message-heavy, a live dashboard is connection-heavy β check which dominates your use.
Managed handles scaling, presence and reconnects for you at a premium. Self-hosting is far cheaper per connection but reconnect storms and horizontal scaling are genuinely hard. For spiky, global apps the managed premium is often worth it.