Perps Latency Benchmark
Network response latency by venue

Live Benchmark Results

Updated -0 measurements
Best post-only p95
-
no matching data
Best taker p50
-
no matching data

Venue Performance

VenueScenarioSubmissionOrderMetricMeasurementsOKCancel p50Cancel p95Cancel p99Cost/runSpeed bumpErrors
Chart controls

Post-only Confirmation

How quickly a resting order is confirmed as placed.

Shaded bands show IQR.
Loading latency data

Batch Post-only Confirmation

Five post-only orders per sample. Native batch venues are labeled separately from manual fanout venues that send concurrent single-order requests.

Shaded bands show IQR.
Loading latency data

Cancel Confirmation

Post-only cleanup cancel latency, measured when the cancel is confirmed through the account feed.

Shaded bands show IQR.
Loading latency data
Loading TPS data

Taker Confirmation

How quickly a marketable order is confirmed, adjusted for published venue delays.

Shaded bands show IQR.
Loading latency data
Loading taker cost data

Benchmark Box

Cloud
AWS EC2
Location
Tokyo (ap-northeast-1)
Availability zone
ap-northeast-1a
Instance
t2.medium

Methodology

Hyperliquid

Post-only BTC order. Confirm latency is measured from completed order-submit write to the matching private account-feed order update. The synchronous exchange response is shown on hover.

Lighter

Post-only BTC order using the maker-only API key. Confirm latency is measured from completed sendTx write to the matching private account order/trade WebSocket event for the client order index. The sendTx queue ack is shown on hover.

Extended

Post-only BTC order. Confirm latency is measured from completed order-submit write to the matching private account order/trade WebSocket event for the external order ID. Batch view labels documented native batch separately from manual fanout.

Nado

Post-only BTC-PERP order. Single-order submit uses Gateway WebSocket and private subscription confirmation. Batch view uses manual fanout over concurrent place_order executes because no native multi-place endpoint is documented.

Aster

Post-only BTCUSDT order using GTX time-in-force. Confirm latency is measured from completed order-submit write to the matching user data stream ORDER_TRADE_UPDATE event for the client order ID. The HTTP submit response is shown on hover.