Switching from QuickNode? Keep your code.
Our streams are wire-compatible with QuickNode's Hyperliquid client stubs: the same hyperliquid.* proto package. Change the endpoint and the key; your stubs, decoders, and everything downstream keep working. You gain the historical archive, derived datasets, and exact-decimal attribution on top.
0+
markets, perps + spot + HIP-3
~$0.0B
traded daily, captured in real time
0.00B+
order events archived
One endpoint. That is the whole diff.
Your generated stubs already speak our wire format. Point the channel at stream.hyperliquidrpc.com:443, swap the key, and every StreamData, L2/L4 book, and block subscription you run today keeps running.
- Auth is an x-api-key entry in gRPC metadata; check your current header name when you swap keys.
- Set max-receive message size to 256 MB or more before subscribing to full book snapshots.
- Streams are long-lived: reconnect with backoff, and treat DATA_LOSS as “fell behind; reconnect and optionally replay from a height.”
- gRPC reflection is enabled, so grpcurl and freshly generated stubs work against the endpoint too.
What changes, beyond the endpoint.
Cells reflect a typical Hyperliquid RPC provider; verify against your current plan. Partial marks are deliberate; precision beats smugness.
| Feature | HyperliquidRPC | QuickNode |
|---|---|---|
| Real-time | ||
| Real-time gRPC streams | ||
| Wallet-level attribution on every fill | ◑ | |
| Exact decimal precision (no floats) | ◑ | |
| Server-side coin / wallet filtering | ◑ | |
| Compatibility | ||
| Works with QuickNode and Dwellir client stubs | Both | ◑ |
| gRPC reflection enabled | ◑ | |
| Historical | ||
| Historical dataset | Full archive | ◑ |
| Derived datasets (OHLCV, PnL, liquidations) | — | |
| Same schema live and historical | — | |
| Platform | ||
| HyperEVM JSON-RPC | ||
| Setup time | Minutes (API key) | Minutes |
Real-time gRPC streams
Wallet-level attribution on every fill
Exact decimal precision (no floats)
Server-side coin / wallet filtering
Works with QuickNode and Dwellir client stubs
gRPC reflection enabled
Historical dataset
Derived datasets (OHLCV, PnL, liquidations)
Same schema live and historical
HyperEVM JSON-RPC
Setup time
◑ = partial or provider-dependent. We keep this table honest: if a cell is out of date, tell us and we will correct it.
Same wire format, more data behind it.
The migration costs one config line. These are the capabilities waiting on the other side of it.
Full wallet attribution
Every fill carries the trader's address, realized PnL, fees, maker/taker role, and position context, not an anonymized tape. Copy-trading and flow analytics become possible.
Exact decimal precision
Prices, sizes, PnL, and fees are byte-exact decimal strings, never floats. Your accounting and backtests reconcile to the chain.
The complete historical archive
1.65B+ order events and 1B+ book updates archived in the same schema as the live stream. Backtest first, deploy with zero remapping.
Both wire formats, one endpoint
We serve the QuickNode-compatible hyperliquid.* package and the Dwellir-compatible gateway side by side. Migrating to us does not lock you into us.
Server-side filtering and replay
Subscribe to just the coins and wallets you need, and replay any stream from a past block height to backfill before going live. Replay starts at Pro.
Derived datasets
OHLCV candles, wallet PnL, liquidation history, and funding-rate series, built from the full tape rather than samples.
Run us side-by-side before you switch.
Wire compatibility makes the evaluation cheap: duplicate your consumer, point the copy at our endpoint, and diff the two feeds against each other. Pro bills month-to-month, so there is no commitment beyond the current period.
Keep QuickNode running
Change nothing in production. Your existing subscription is the control group.
Point a duplicate client at us
Same stubs, same request messages; only the channel target and the x-api-key metadata differ. A Free key covers trades, the BTC ticker, and the L2 book on up to 2 coins.
Diff the feeds, then decide
Compare message contents, gaps, and latency over a few days of real trading. Switch when the data says so, not because a landing page did.
The full migration guide
Endpoint swap, auth metadata, message-size limits, DATA_LOSS semantics, and the gotchas in between, documented step by step.
The questions migrations actually raise.
Mechanics, gotchas, and tier gates, answered plainly. Anything missing lives in the migration guide.
No. We serve the same hyperliquid.* proto package your QuickNode client was generated against. If you prefer fresh stubs, reflection is enabled; generate directly from the endpoint or from our published .proto files.
Two values: the channel target becomes stream.hyperliquidrpc.com:443, and your API key moves into x-api-key gRPC metadata. Request messages, stream types, and response decoding are unchanged.
Yes. Run a duplicate consumer against our endpoint and diff it against your existing feed. Every stream is on both plans, and Pro bills month-to-month so you can stop after one period if the data does not fit.
The wire format is identical; the operational envelope is documented rather than assumed. Set a ≥256 MB max-receive size for book snapshots, reconnect with backoff, and read our DATA_LOSS semantics before going live. The migration guide lists every gotcha we know about.
The full L4 order book and raw book diffs are on both plans; replay from a block height is Ultra. The complete gate-by-gate breakdown is on the pricing page.
Change one endpoint. Keep everything else.
A Free key takes minutes and is sized for a side-by-side evaluation against your current feed.