Overview
Hyperliquid Data Stream is a low-latency WebSocket data service built on top of a Hyperliquid validator node. It provides real-time access to the node-level market and execution data through a single WebSocket connection.
Clients can use the stream to receive:
- Blocks applied by the node
- Per-coin order book diffs
- Mempool transactions, when enabled
- Timing metrics
- Low-latency ping frames
- Server-side error events
The service is designed for teams that need fast and reliable Hyperliquid data for execution systems, trading infrastructure, monitoring, research, and latency-sensitive workflows.
How it works
Hyperliquid Data Stream runs as a WebSocket service in front of a Hyperliquid node. Clients connect to the provided endpoint, subscribe to the required data streams, and receive real-time updates as soon as they are available from the node.
The service supports two modes:
| Mode | Purpose | Recommended use |
|---|---|---|
| JSON mode | Simple text-based order book diffs | Quick testing and inspection |
| Binary mode | Blocks, orders, metrics, pings, and mempool data | Production integrations |
JSON mode is useful for a quick smoke test. For production, use binary mode.
Endpoint
Hyperliquid Data Stream speaks WebSocket over plain TCP.
ws://<host>:<port>The endpoint URL is provided during onboarding.
Important: The service uses ws://, not wss:/.
Access control
Access is granted per source IP at the network layer.
If a client’s egress IP changes, the new IP must be added to the allowlist. Otherwise, the TCP connection may hang or time out.
Limits
| Limit | Value | What happens if exceeded |
|---|---|---|
| Connections per source IP | 4 | A new connection is accepted and then immediately closed |
| Read rate | See Backpressure section | Slow clients are disconnected |
What's next?
Updated 3 days ago