API Documentation

Build one integration for the documented Hyperliquid and Lighter market-data routes. Keep venue routing, response parsing, and coverage checks in one place instead of separate collectors for each venue; REST and WebSocket responses remain route-specific, and Parquet handles bulk jobs.

Market Data API

One integration for market data.

Build with Hyperliquid L4 and Lighter L3 order-level data, trades, candles, funding, open interest, and liquidations over REST, WebSocket, and Parquet.

First call

Create a key and confirm one authenticated call

Create a key in the dashboard, then run the bounded request below and inspect its response.

Create a key and run one call →

Make your first call

Set OXARCHIVE_API_KEY, then make one small authenticated request. The request uses the native Hyperliquid L2 route and asks for one price level per side.

export OXARCHIVE_API_KEY="0xa_your_api_key"
curl "https://api.0xarchive.io/v1/hyperliquid/orderbook/BTC?depth=1" \
  -H "X-API-Key: $OXARCHIVE_API_KEY"

A successful 200 response has success: true, an order-book object in data, and a UUID-shaped meta.request_id:

{
  "success": true,
  "data": {
    "coin": "BTC",
    "symbol": "BTC",
    "timestamp": "2026-08-28T00:00:00.000Z",
    "bids": [{ "px": "61077.0", "sz": "0.0003", "n": 1 }],
    "asks": [{ "px": "61078.0", "sz": "11.73086", "n": 1 }],
    "mid_price": "61077.5"
  },
  "meta": {
    "request_id": "550e8400-e29b-41d4-a716-446655440000"
  }
}

The values above illustrate the response shape. Keep meta.request_id or the x-request-id response header with the route and symbol. Quickstart covers the same call with language examples, REST authentication covers HTTP key mechanics, and Errors and request IDs covers recovery.

Choose the job

Historical pulls

Use REST for historical pulls, bounded windows, point-in-time snapshots, and route-specific parameters. Check Venue coverage before choosing a symbol and window. For the Hyperliquid core route map, open Hyperliquid historical data API before choosing a REST, replay, or export surface.

Real-time streams and replay

Use WebSocket for real-time channels. Replay is channel-specific: standard channels use timed replay, core Hyperliquid L4 uses checkpoint-anchored bulk delivery, and Spot, HIP-3, and HIP-4 L4 channels are real-time only. Check WebSocket channels before choosing a mode.

Bulk research files

Use Data Catalog exports when the job needs Parquet files instead of request-by-request API access. Keep the venue, market type, schema, symbol set, time window, and rights decision with the export.

Choose a venue, market type, and depth

0xArchive covers Hyperliquid and Lighter. Venues and market types explains how core perpetuals, Spot, HIP-3, HIP-4, and Lighter markets map to API namespaces. See Get market data to choose between API access and Parquet exports. The strongest order-level paths are different by venue:

Workload	Data grain	Start here
Hyperliquid L4	Resting orders across served levels, with diffs and checkpoint-based reconstruction	L4 order book
Lighter L3	Individual resting orders with Lighter order indexes; up to 250 orders per side from March 5, 2026	L3 order book
Aggregated L2	Price levels with `px`, `sz`, and `n`; native Hyperliquid L2 is capped at 20 levels per side	L2 order book

Keep L4 on Hyperliquid and L3 on Lighter. Order-level books are separate from fills, order lifecycle, order flow, and TP/SL routes.

Check coverage and quality

Before a backtest, alert, export, or model run depends on a result, check the exact market, data type, and window on Venue coverage. Coverage is checked per symbol and data type. Use Data quality for freshness, incidents, and gaps. The /v1/symbols coverage_by_type data and the quality routes provide the checks for a selected symbol; a venue-level summary is not a symbol-level result. Review Rate limits for capacity and Data rights before exporting or shipping results.

Choose an interface

Interface	Use it for	Credential or boundary
REST API	Bounded history, snapshots, and route lookups	Send `X-API-Key` where the endpoint requires it
WebSocket	Real-time subscriptions and documented replay modes	Send an `Authorization` bearer header during the opening handshake
SDKs, CLI, Skill	API-key-backed client workflows	Load the key from the runtime environment or secret store
Hosted MCP server	Tool calls in MCP-capable clients	Client-managed OAuth; no 0xArchive API key
Data Catalog	Bulk Parquet delivery	Use the dashboard account, quote, checkout, and job flow
OpenAPI	Exact paths, parameters, and security for code generation	Use human pages for venue choice, recovery, coverage, and response fields

For agent builders and quant teams

Agent builders can use OpenAPI for exact paths, parameters, and security, then use route pages and the human guides for response fields where they are enumerated, venue-family choice, replay mode, recovery, and data-quality decisions. Quant teams should keep venue family, symbol, data type, window, depth, quality state, and request ID beside each result.

Run the Quickstart first. Then choose Venue coverage, WebSocket channels, or Data Catalog based on the job.

Get started

Plans & limits

Make your first call

Choose the job

Historical pulls

Real-time streams and replay

Bulk research files

Choose a venue, market type, and depth

Check coverage and quality

Choose an interface

For agent builders and quant teams

Next

​Make your first call

​Choose the job

​Historical pulls

​Real-time streams and replay

​Bulk research files

​Choose a venue, market type, and depth

​Check coverage and quality

​Choose an interface

​For agent builders and quant teams

​Next

Make your first call

Choose the job

Historical pulls

Real-time streams and replay

Bulk research files

Choose a venue, market type, and depth

Check coverage and quality

Choose an interface

For agent builders and quant teams

Next