# Hyperliquid API Rate Limits

> The Hyperliquid IP limit of 1200 weight each minute, the weight of each request, the address limit of 1 request for each 1 USDC traded, and WebSocket and open-order limits.

- Canonical URL: https://hiperwire.io/build/hyperliquid-api-rate-limits
- Last verified: November 17, 2026
- Publisher: HIPERWIRE (https://hiperwire.io). Not affiliated with Hyperliquid.

Hyperliquid has two separate rate limits. One counts requests from your IP address. The other counts trading actions from your account address. A bot can hit either one.

## Summary

| Limit | Applies to | Budget |
| --- | --- | --- |
| IP limit | All REST requests from one IP address | 1200 weight each minute |
| Address limit | Trading actions of one account | 10,000 requests, plus 1 for each 1 USDC traded |
| WebSocket | Connections from one IP address | 10 connections, 1000 subscriptions |
| Open orders | One account | 1000 orders, more with volume |

## IP limit: 1200 weight each minute

Each REST request has a weight. All requests from one IP address share 1200 weight each minute.

### Info requests

| Weight | Request types |
| --- | --- |
| 2 | `l2Book`, `allMids`, `clearinghouseState`, `orderStatus`, `spotClearinghouseState`, `exchangeStatus` |
| 20 | Most other info requests, for example `meta`, `metaAndAssetCtxs`, `openOrders`, `userFills` |
| 60 | `userRole` |
| 40 | All explorer requests |

Some requests cost more when they return more data:

- `recentTrades`, `historicalOrders`, `userFills`, `userFillsByTime`, `fundingHistory`, `userFunding` and similar list requests add weight for each 20 items returned.
- `candleSnapshot` adds weight for each 60 items returned.

At weight 2, you can poll `allMids` or `l2Book` 600 times each minute. At weight 20, you can call `metaAndAssetCtxs` 60 times each minute. A loop that calls three weight-20 requests each second uses 3600 weight each minute and is blocked.

### Exchange actions

An exchange action has a weight of `1 + floor(batch_length / 40)`.

- One order has a weight of 1.
- One request with 39 orders has a weight of 1.
- One request with 79 orders has a weight of 2.

Batching is the main way to stay inside the IP limit. In the Python SDK, use `bulk_orders` and `bulk_cancel`.

## Address limit: 1 request for each 1 USDC traded

This limit counts actions, not info requests. It applies to each account. A sub-account counts as a separate account.

- Each address starts with a budget of 10,000 requests.
- Each 1 USDC of traded volume adds 1 request to the budget. The count is cumulative from the first trade of the address.
- When the budget is used, the address can send 1 request each 10 seconds.
- A batch of `n` orders counts as `n` requests for this limit. It counts as 1 request for the IP limit.

Cancels have a higher limit, so a rate-limited account can still remove its orders. The cancel limit is `min(limit + 100000, limit * 2)`, where `limit` is the normal budget.

This limit affects bots that place and cancel many orders but trade little volume. A market maker that sends 50,000 orders and fills $20,000 of volume has used 50,000 requests against a budget of 30,000.

### Check your budget

```python
from hyperliquid.info import Info
from hyperliquid.utils import constants

info = Info(constants.MAINNET_API_URL, skip_ws=True)
limit = info.post("/info", {"type": "userRateLimit", "user": "0xYourMainAccount"})
# {'cumVlm': '0.0', 'nRequestsUsed': 0, 'nRequestsCap': 10000, 'nRequestsSurplus': 0}
```

`nRequestsCap - nRequestsUsed` is the number of requests you have left. Pass the main account address, not the API wallet address.

## WebSocket limits

These limits apply to one IP address:

- 10 connections.
- 30 new connections each minute.
- 1000 subscriptions.
- 10 unique users across user-specific subscriptions.
- 2000 messages sent to Hyperliquid each minute, across all connections.
- 100 simultaneous in-flight post messages, across all connections.

Use a WebSocket subscription for data that changes often, such as the order book, trades and your own fills. A subscription does not use REST weight.

## Open order limit

An account can have 1000 open orders. The limit increases by 1 order for each 5 million USDC of volume, to a maximum of 5000.

When an account has 1000 or more open orders, Hyperliquid rejects new reduce-only orders and new trigger orders.

## HyperEVM RPC limit

The public RPC endpoint `rpc.hyperliquid.xyz/evm` allows 100 requests each minute. Other RPC providers set their own limits.

## What to do when you are rate limited

- **Batch orders and cancels.** The Hyperliquid docs suggest one batch each 0.1 seconds.
- **Read with WebSocket, not with polling.** Subscribe to `l2Book`, `trades` and `userFills`.
- **Do not send a cancel again when you already have its result.** Each cancel uses budget.
- **Cache `meta`.** Asset names and `szDecimals` change rarely. Read them one time at startup.
- **Use weight-2 requests where you can.** `clearinghouseState` (weight 2) gives positions. `allMids` (weight 2) gives all mid prices in one request.

If the address limit stops you, the account must trade more volume to earn budget. Until then, the address can still send 1 request each 10 seconds.

## Runnable example

[examples/04_rate_limit.py](https://github.com/atlasdetitan/hyperliquid-trading-bots/blob/master/examples/04_rate_limit.py) prints the request budget of any address.

## Sources

- [Rate limits and user limits](https://hyperliquid.gitbook.io/hyperliquid-docs/for-developers/api/rate-limits-and-user-limits), Hyperliquid docs
- [Nonces and API wallets](https://hyperliquid.gitbook.io/hyperliquid-docs/for-developers/api/nonces-and-api-wallets), Hyperliquid docs, for the batch interval
- The `userRateLimit` response was read from the live API on November 17, 2026.
