Rate Limits

Rate limits protect the trading gateway and apply independently across REST request weight, order placement, WebSocket usage, and user-address actions. Track every applicable dimension instead of relying on a single request counter.

Integration guidance

• Budget request weight before sending high-volume queries or batches.

• Apply bounded exponential backoff when a request is rate limited.

• Reconcile remote state before retrying an action whose result is unknown.

• Use WebSocket streams for realtime state changes and REST for snapshots and recovery.

• Distribute concurrent trading processes across dedicated API keys to avoid shared nonce contention.

IP-Based Rate Limits

All REST API requests consume weight from a fixed 1-minute window. Once the total weight consumed within the window reaches the limit, subsequent requests are rejected until the window resets.

Endpoints not listed below default to weight 20.

Overall Limits

Dimension	Limit
Weight per minute per IP	1200

---

Public Endpoints

Endpoint	Weight	Notes
Query announcements	5	GET /api/v1/announcements; mainnet only
Query announcement detail	5	GET /api/v1/announcements/detail/{id}; mainnet only
Query user rate limits	20	GET /api/v1/user/{userAddress}/ratelimit

Deposit Address Endpoints

Endpoint	Weight	Notes
Create one deposit address	200	POST /api/v1/user/{userAddress}/deposit-address
Create all supported deposit addresses	200	POST /api/v1/user/{userAddress}/deposit-addresses; charged per request, not per returned address

Spot Market Endpoints

Endpoint	Weight	Notes
Query symbols	2
Query coins	2
Query tickers	2
Query mini tickers	2
Query book tickers	2
Query order book	5 / 10 / 20	depth ≤100 → 5, 101–500 → 10, >500 → 20
Query candles/klines	20
Query recent trades	20

Spot Accounts Endpoints

Endpoint	Weight	Notes
Query balances	5
Query open orders	5
Query state for frontend	5
Query API Keys	5
Query fee rate	2
Query order history	20	+ 1 extra per 20 items returned
Query user trades	20	+ 1 extra per 20 items returned
Transfer asset to EVM or perps	10

Spot Trading Endpoints

Endpoint	Weight	Notes
Place multiple orders	1 + floor(N/40)	N = number of orders
Cancel multiple orders	1 + floor(N/40)	N = number of orders
Replace multiple orders	1 + floor(N/40)	N = number of orders
Schedule cancel orders	1

---

Perps Market Endpoints

Endpoint	Weight	Notes
Query symbols	2
Query coins	2
Query tickers	2
Query mini tickers	2
Query mark prices	2
Query book tickers	2
Query order book	5 / 10 / 20	depth ≤100 → 5, 101–500 → 10, >500 → 20
Query candles/klines	20
Query recent trades	20

Perps Account Endpoints

Endpoint	Weight	Notes
Query balances	5
Query open orders	5
Query open positions	5
Query state for frontend	5
Query API Keys	5
Query fee rate	2
Query order history	20	+ 1 extra per 20 items returned
Query position history	20	+ 1 extra per 20 items returned
Query trades	20	+ 1 extra per 20 items returned
Query funding history	20	+ 1 extra per 20 items returned
Transfer asset to spot	10

Perps Trading Endpoints

Endpoint	Weight	Notes
Place multiple orders	1 + floor(N/40)	N = number of orders
Cancel multiple orders	1 + floor(N/40)	N = number of orders
Replace multiple orders	1 + floor(N/40)	N = number of orders
Modify TP/SL order	1
Schedule cancel orders	1
Update leverage	1
Update isolated margin	1

---

Dynamic Weight Rules

Orderbook Depth

Depth (limit)	Weight
≤ 100 (default)	5
101 – 500	10
> 500	20

Batch Order Formula

weight = 1 + floor(batch_size / 40)

Batch size	Weight
1 – 39	1
40 – 79	2
80 – 119	3

History Extra Weight

History endpoints incur additional weight after the response, based on items returned:

extra_weight = floor(items_returned / 20)

Kline Cache Miss Extra Weight

Kline requests have a base weight of 20. Additional weight may apply based on the number of rows returned:

extra_weight = max(1, floor(rows_returned / 25))

---

Order Placement Limits

Separate from the weight budget:

Client type	Limit	Scope
API key	600 orders / minute and 20 orders / second	per account
Web (no key)	60 orders / minute	per account

For batched requests, the number of orders counted against these limits is the batch length: order_count = len(batch).

---

WebSocket Limits

Separate from the REST weight budget:

Limit	Value
Concurrent connections per IP	10
New connections per IP per minute	30
Subscriptions per IP (all connections)	1000
Unique users per IP	10
Messages per IP per minute	2000
Messages per connection per minute	2000
Inflight requests per IP	100

---

Address-Based Rate Limits

Address-based limits apply per user address and only affect actions (order placement, cancellation, etc.), not info/query requests.

The rate limiting logic allows 1 request per 1 USDC traded cumulatively since address inception. For example, with an order value of 100 USDC, this requires a fill rate of 1%. Each address starts with an initial buffer of 10,000 requests.

When rate limited, an address is allowed one request every 10 seconds.

Cancels have a higher cumulative limit to ensure open orders can always be wound down:

cancel_limit = min(limit + 100000, limit * 2)

where limit is the default limit for other actions.

Each user has a default open order limit of 1000 across all symbols combined (spot + perps).

---

Batched Requests

A batched request containing N orders (or cancels) is counted differently depending on the rate-limit dimension:

Dimension	Counted as
IP-based weight	1 request (weight per the batch formula above)
Address-based limit	N requests (one per order/cancel in the batch)