0xArchive is a query API for Hyperliquid and Lighter market data: order books at a timestamp, trades, candles, funding and open interest over REST and WebSocket. For many jobs it is a good fit. We compare it with Dwellir in 0xArchive vs Dwellir for Hyperliquid data.
This page lists the alternatives and the job each one fits. Dwellir is one of them, and we wrote this page. Every date and claim comes from the provider's own public page, checked on 11 November 2026. Prices change often, so follow each link for current terms. Dwellir's prices, as of 11 November 2026, are in its section below.
Alternatives at a glance
| Option | What you get | History from | Access | Fits |
|---|---|---|---|---|
| 0xArchive (baseline) | Book at a timestamp, trades, candles, funding, open interest, L4; Hyperliquid and Lighter | Core trades and book from 15 Apr 2023; funding and OI from 20 May 2023 | REST, WebSocket, Parquet exports | One-call answers and long market history |
| Hyperliquid S3 buckets and Info API | L2 book snapshots, asset contexts, node fills, replica_cmds, misc events | Varies by bucket | S3, requester pays; Info API | Free raw material, if the monthly hyperliquid-archive uploads are enough |
| Your own non-validating node | Everything the node writes, from the day you start | Your start date | Local disk | Teams that want full control going forward |
| Dwellir | Raw node files, gRPC and WebSocket streams, fills and candle APIs, RPC and HyperEVM; free plan, paid API plans from $49 a month | replica_cmds from 25 Jan 2025 | S3 key, gRPC, WebSocket, REST | Replay, audit, order lifecycle, node infrastructure |
| Tardis.dev | Recorded Hyperliquid WebSocket channels: trades, l2Book, bbo, fastBook, activeAssetCtx (funding, open interest, mark) | 29 Oct 2024 | Replay API, CSV downloads | Replay of the public feed: tick-level trades, book snapshots every few seconds |
| Hydromancer | L2 and L4 order book streams with addresses, fill history, 1-second candles, Parquet storage | Fills from 28 Jul 2025 | API, streams | Live data plus recent history from one vendor |
Sources: Hyperliquid historical data docs, Dwellir historical data docs, Tardis.dev Hyperliquid details, Hydromancer. All read on 11 November 2026.
1. Hyperliquid's own S3 buckets and Info API
Hyperliquid publishes historical data itself. The hyperliquid-archive bucket holds L2 book snapshots under market_data and asset contexts under asset_ctxs. The hl-mainnet-node-data bucket holds node fills (node_fills_by_block, with older formats in node_fills and node_trades), explorer_blocks, replica_cmds, and misc_events_by_block for transfers, staking and funding.
The limits are stated in the same docs. The requester pays transfer costs. The hyperliquid-archive market data bucket is uploaded "approximately once a month", with no guarantee of timely updates, and data may be missing. No candles or spot asset data are provided through S3. The Info API fills some gaps for recent data, but its candle endpoint returns only the most recent 5,000 candles.
It fits teams that are comfortable with LZ4 files and AWS requester-pays, and can work with a monthly update for the hyperliquid-archive market data.
2. Run your own non-validating node
A Hyperliquid node can write fills, order statuses, raw book diffs, misc events and more to disk as it runs. You get every record from your start date onward, at the resolution the node produces, and you own the storage.
The cost is hardware, disk and operations. The output grows fast: 0xArchive's peering docs note that writing order statuses alone runs to several hundred GB a day. A node also has no history before it started, so it suits forward collection, not backfill. A reliable upstream peer matters too; both 0xArchive and Dwellir sell Tokyo peering for this. Our peering guide covers the setup.
It fits teams that already run infrastructure and want full control from now on.
3. Dwellir
Dwellir runs Hyperliquid nodes and sells the files they write, so you do not have to run one. The S3 archive includes replica_cmds from 25 January 2025, fills from 22 March 2025, order statuses and raw book diffs from 22 January 2026, periodic ABCI state plus generated order book snapshots from 14 April 2026, misc events from 29 June 2026 and TWAP statuses from 1 July 2026. The coverage table lists the size and file count of each prefix.
For processed data, the Hyperliquid Index serves fills and 1s, 1m and 5m candles from 27 July 2025 over an API, and full-history candle exports are self-serve in CSV or Parquet. For live data, there is gRPC streaming of raw node feeds and the Order Book WebSocket with L2 up to 100 levels a side and L4 with user addresses. Within the gRPC retention window, GetOrderbookSnapshot returns the order book snapshot covering a timestamp in one call. Dwellir also runs Hyperliquid JSON-RPC, HyperEVM and dedicated nodes.
What Dwellir does not have: funding and open interest series, history before 2025, Lighter, or a REST call that returns the book at any past timestamp. Archive datasets are quoted by email.
On 11 November 2026, API plans started at $49 a month, the Order Book WebSocket at $199 a month plus usage, gRPC at $299 a month plus usage, and peering at $500 a month per slot (pricing, Hyperliquid).
It fits chain replay, order lifecycle and queue simulation, audit work, bulk transfer into your own storage, and teams that also need node infrastructure.
4. Tardis.dev
Tardis.dev records exchange WebSocket feeds and replays them. Its Hyperliquid data starts on 29 October 2024 and includes trades, l2Book (20-level snapshots, observed about every 5.4 seconds), bbo (from 26 June 2025), fastBook (5 levels, from 17 June 2026), and activeAssetCtx and activeSpotAssetCtx (funding, open interest and mark price). Trades recorded from 1 December 2024 include buyer and seller addresses, and HIP-3 markets are covered from 2 December 2025. CSV files for the first day of each month can be downloaded without an API key.
Tardis.dev captures the public WebSocket feed, not node output, so it has no order statuses or replica_cmds. It also offers the data in a normalized CSV format shared across many other exchanges, which helps if you backtest across venues.
It fits replay of the public Hyperliquid feed, with tick-level trades and book snapshots every few seconds, especially next to other exchanges.
5. Hydromancer
Hydromancer sells Hyperliquid data streams and enhanced endpoints. Its site lists L2 and L4 order book streaming with address visibility, historical data and market snapshots in Parquet at 1-second resolution, every fill since 28 July 2025, and 1-second candles.
Hydromancer also publishes Reservoir, a free archive of fills, 1-second candles, daily snapshots and order books from 28 July 2025, updated daily on requester-pays S3, with no plan or API key needed. Its site lists history from 28 July 2025 and does not publish an earlier start. It fits teams that want live streams and recent history from one vendor, without running a node.
Which one to pick
| Your job | Start with |
|---|---|
| Book at a timestamp in one request, funding and open interest from 2023 | 0xArchive |
| Free raw files, no rush | Hyperliquid S3 buckets |
| Exact node records for replay or audit, order lifecycle, bulk download | Dwellir |
| Public feed replay across many exchanges | Tardis.dev |
| Forward collection under your own control | Your own node, with a reliable peer |
| Live L2 and L4 streams | 0xArchive, Dwellir or Hydromancer; test each from your own servers |
You can also combine two: a query API for research and raw files for the windows that need them. For a longer list of vendors with coverage dates and billing models, see Hyperliquid backtesting data: 20 sources compared.
To try Dwellir's archive, email support@dwellir.com with the prefix and date range and ask for a sample file, or create a free account to call the Index APIs.
