fxapis, measured. Not promised.
Three numbers describe how fxapis feels in use: how long an account takes to come online, how long an order takes to reach the broker and back, and how close together orders leave when you trade many accounts at once. Here is each one, how we measured it, and what changes it in the real world.
From asking for an account to come online until it is signed in at the broker and ready to trade. One run.
From our servers sending an order to the broker's answer, to a broker with trade servers in London.
Between the first and the last of 100 accounts' orders leaving fxapis, in one multi-account order, in a local benchmark against an instant broker.
9.6 seconds.
What: the time from POST /v1/accounts/prepare to the account's status reading ready — signed in at the broker, prices arriving, orders accepted.
Where and against what: our production service in the EU (Amsterdam), with a retail broker's MetaTrader 5 demo account whose trade servers are in London, on 29 September 2026. This figure is from one run.
What it means for you: an account in on-demand mode that has gone offline — after 15 quiet minutes — needs about this long before its first order. That is why click-to-trade prepares a member's account when they open a signal, not when they approve it, and why accounts that must react instantly should be always connected. Once an account is online, orders go straight through.
Sign-in time also depends on the broker's own servers. A busy broker can take longer, which is why the polling guide suggests a generous timeout.
96–110 milliseconds.
What: the time from our server sending a market order to the broker's trade server returning its answer, for an account that was already online.
Where and against what: the same production service and demo account, on 29 September 2026. Our servers are in Amsterdam; the broker's trade servers are in London. The figure includes the broker's own time to process the order.
- 01Your backendAnywhere
- 02fxapis APIEU · Amsterdam
- 03MT5 terminalOur cloud, same region
- 04Broker trade serverWherever your broker runs it
30 ms across 100 accounts.
What: in a multi-account order, the time between the first account's order being sent and the last one's, measured from each order's own send time. Every multi-account order reports this figure as dispatchSpreadMs, so you can watch it on your own trades.
Against what: a local broker that answers instantly. That removes the broker's latency — which differs by broker and by moment — and leaves the part fxapis controls: how close together it sends the orders.
What it shows: the cost per account falls as the order grows, which is what sending in parallel should look like. Real brokers add their own execution time to each account, so fills will still arrive at slightly different moments and prices.
| Aspect | First to last | Per account |
|---|---|---|
| 10 accounts | 17 ms | 1.70 ms |
| 50 accounts | 26 ms | 0.52 ms |
| 100 accounts | 30 ms | 0.30 ms |
What changes your numbers.
- Your broker. Where its trade servers are, and how it executes orders, matter more than anything else. A broker in New York or Asia will be further from our EU servers than one in London.
- Whether the account is online. An offline account signs in first — about 10 seconds. Keep time-critical accounts always connected, or prepare them ahead.
- The market. Around news, prices move quickly; brokers requote or reject orders whose price has moved beyond the deviation you allow.
- Your own servers. The time from your backend to our API is on top of everything here. Servers in or near the EU keep it small.
- How many accounts at once. Larger multi-account orders spread slightly wider, but each extra account adds less.
About these numbers
Are these numbers guaranteed?
No. They are measurements of our own service under the conditions described on this page. Your results depend on your broker, where its trade servers are, the market at the time and where your own servers are. Service levels are only agreed in an Enterprise contract.
Why is the multi-account spread measured against an instant broker?
To isolate our own contribution. A real broker adds its own execution time to every order, and that time varies from broker to broker and moment to moment. Measuring against a broker that answers instantly shows how far apart fxapis sends the orders, which is the part we control.
Why is the round trip about 100 milliseconds?
It is the time from our servers sending an order to the broker's trade server answering, and it includes the broker's own processing. Our servers are in Amsterdam; the broker we measured keeps its MT5 trade servers in London. A broker with servers elsewhere, or a slower execution model, will show a different number.
How do I get the fastest execution?
Keep accounts that must act instantly in always-connected mode, or bring them online ahead of time with POST /v1/accounts/prepare, so no order waits for a sign-in. Run your own backend close to our EU servers, and set a sensible price deviation so fast markets don't cause needless rejections.
Do you publish uptime figures?
Not yet. We will publish uptime once there is a long enough record to make the figure meaningful, and not before.
See how the numbers shape a product: click-to-trade for signal providers and copy trading on multi-account orders.
Measure it on your own broker.
Connect a demo account at your broker and time your first orders yourself.