# How to get funding rates and open interest

Both numbers are free from the venue and one call away from an aggregator. But no two exchanges quote them in the same units or on the same clock.

*https://cryptomarkets.tools/how-to/get-funding-and-open-interest · next to Crypto Market Data APIs*

**Answer:** Take them from the venue if one book is the question: Binance, Bybit and OKX all publish funding history and open interest on unauthenticated endpoints. Take them from Coinalyze free, CoinGlass paid or Velo already rescaled if you need every venue at once. Whichever you pick, nothing is comparable across venues until you convert it to one interval and one unit.

## Approaches

*In the author’s order. Paid placement does not affect it.*

1. [Coinalyze](https://cryptomarkets.tools/tools/coinalyze.md) — Free key at 40 calls a minute over sixteen venues, with current, predicted and historical funding split into three endpoints, against a short rolling window.
2. [CoinGlass API](https://cryptomarkets.tools/tools/coinglass.md) — Funding and open-interest candles across thirty futures venues, aggregated or per pair, on every plan from 35 dollars a month.
3. [Velo](https://cryptomarkets.tools/tools/velo.md) — Cross-venue funding normalised to a per-eight-hour figure, which is the rescaling most products leave to you, with open interest broken out by venue.
4. [Binance Market Data API](https://cryptomarkets.tools/tools/binance-api.md) — The venue's own funding history and open-interest statistics, free and unkeyed, with a flat-file funding archive underneath the REST surface.
5. [Bybit Market Data API](https://cryptomarkets.tools/tools/bybit-api.md) — Same shape, and it publishes the funding interval per symbol in minutes — unless your IP is in the US or mainland China, which gets a 403.

## The short way

Ask the venue. On the largest perpetual books both numbers are public, and neither endpoint wants a
key, an account or a funded balance.

```text
GET https://fapi.binance.com/fapi/v1/fundingRate?symbol=BTCUSDT&limit=100
GET https://fapi.binance.com/futures/data/openInterestHist?symbol=BTCUSDT&period=5m
GET https://api.bybit.com/v5/market/funding/history?category=linear&symbol=BTCUSDT
GET https://api.bybit.com/v5/market/open-interest?category=linear&symbol=BTCUSDT&intervalTime=5min
```

[Binance's market data API](https://cryptomarkets.tools/tools/binance-api) returns settled funding rates with the millisecond
timestamp each one settled at, and an open-interest series at periods from five minutes upward.
[Bybit's](https://cryptomarkets.tools/tools/bybit-api) is the same shape under different parameter names. Both meter you per
IP rather than per key, so containers behind one egress address share one budget, and Bybit returns
403 to any request from a US or mainland China address — which decides the question before any
feature does.

If you want more than one venue, do not write that loop first. [Coinalyze](https://cryptomarkets.tools/tools/coinalyze) gives
you sixteen venues of funding, predicted funding, open interest and liquidations through a free key
at 40 calls a minute. That is the whole short way for most people. The rest of this page is about
why the numbers that come back cannot simply be put in the same column.

## What the options are

**One venue, direct.** Binance, Bybit and [OKX](https://cryptomarkets.tools/tools/okx-api) each publish the funding rate for
the coming settlement, the settled funding history and open interest without authentication. Use
them when the question is about one book — this venue's positioning, this venue's payments — and
accept three costs: rate limits counted per IP, terms you have to read before showing anyone the
result, and a normalisation job that is now yours rather than a vendor's.

**Every venue at once, free.** Coinalyze is a futures screener with an API attached, and the API is
on the free tier along with everything else. Twelve endpoints cover supported exchanges, markets,
current and historical open interest, current, predicted and historical funding, liquidations, the
long/short ratio and OHLCV. The catch is memory: the documentation states that 1,500 to 2,000
datapoints are kept for any intraday granularity from one minute to twelve hours and the old ones
are deleted each day, so at one-minute resolution that is somewhere between 25 and 33 hours of
history, permanently. Daily bars are kept.

**Every venue at once, paid.** The [CoinGlass API](https://cryptomarkets.tools/tools/coinglass) is the widest list in this
catalogue — thirty futures venues, including dYdX, Hyperliquid, Drift, Paradex and Lighter
alongside the usual names. Funding and open interest arrive as OHLC candles per pair and aggregated
across exchanges, with stablecoin-margined and coin-margined kept apart rather than summed, and
those candle endpoints are on every plan. What the plan ladder buys is resolution and licence: the
35-dollar tier offers nothing finer than a four-hour bar, minute bars start at the 379-dollar tier,
and the two cheapest plans are marked personal use.

**Already rescaled.** [Velo](https://cryptomarkets.tools/tools/velo) is a derivatives terminal with the dataset behind it sold
in the same subscription, and its funding rates are normalised to a per-eight-hour figure across
venues — the one conversion most products leave to the reader. Open interest is broken down by the
venue it came from rather than merged, which is the difference between "open interest rose" and
"open interest rose on Bybit while Binance was flat". Read the billing terms first: on monthly
billing the archive is three months deep, and the full five-plus years needs the annual plan.

## Where this breaks

**The settlement interval is a property of the symbol, not of the market.** OKX's own explainer
says the fee is charged or paid every 8 hours, at 00:00, 08:00 and 16:00 UTC, "by default, unless
specified otherwise (i.e. every 1, 2 or 4 hours)". Bybit publishes `fundingInterval` per instrument
and documents its unit as minutes, and its funding-history page warns outright that each symbol has
a different funding interval. Binance keeps a separate endpoint whose entire job is to list the
symbols whose `fundingIntervalHours` has been adjusted away from the default. Hyperliquid pays
every hour: its documentation states that the formula produces an eight-hour rate and that one
eighth of it is paid hourly, capped at 4% an hour. Two venues quoting "0.01%" are quoting the same
string and two different quantities, and neither the [funding rate](https://cryptomarkets.tools/glossary/funding-rate) itself
nor any chart of it tells you which.

**So nothing is comparable until you rescale it**, and the rescaling needs a number you have to
fetch separately:

```python
# The venue quotes a rate per settlement. Convert before putting two venues side by side.
rate = 0.0001            # one settlement's rate, exactly as the endpoint returned it
interval_hours = 8       # Binance fundingIntervalHours; Bybit fundingInterval / 60; 1 on Hyperliquid

per_8h = rate * (8 / interval_hours)
annualised = rate * (24 / interval_hours) * 365
```

That annualised figure is a convention, not a fact: it assumes the rate persists and it ignores
compounding. Write down which convention you used, because the same series published two ways is
how a chart ends up disagreeing with itself.

**Predicted funding and realised funding are two different series.** The rate a ticker shows
belongs to a settlement that has not happened and keeps being recalculated until it does — OKX
states that the fee is assessed using the funding rate calculated in the minute before assessment.
The settled number is the one the funding-history endpoint returns, stamped with the time it
settled. Coinalyze ships current, predicted and historical funding as three separate endpoints,
which is the same distinction admitted in a product surface. Storing a predicted rate in the row
where a realised one belongs is a silent error: it never throws, it just leaves you with a history
nobody else can reproduce.

**Open interest is quoted in three different units, and only some of the pages say which.** Bybit
documents the unit of its open-interest figure per contract type in one line: BTCUSD, the inverse
contract, is in USD, and BTCUSDT, the linear one, is in BTC. Binance's COIN-M open-interest
statistics annotate `sumOpenInterest` as `unit: cont` and `sumOpenInterestValue` as `unit: base
asset` — so the second field is a quantity of coin, not a dollar figure — while the USDⓈ-M page
carries the same two field names with no unit annotation at all. And a count of contracts only
becomes a quantity through the contract size, which is published per instrument in exchange
information and is not the same number on every symbol. Three denominations, one metric name; see
[open interest](https://cryptomarkets.tools/glossary/open-interest) for what the metric itself does and does not count.

**A cross-venue total is a sum of numbers defined differently.** Every aggregate figure is the
result of a conversion per instrument, a choice of denominator, and a decision about which venues
are in the population — Coinalyze's sixteen, CoinGlass's thirty and Velo's six are three different
markets before anyone converts anything. Two smaller traps sit inside that. Coinalyze lists
tokenized equity and commodity derivatives in the same table, marked "(Derivatives)", so a total
read off the top of it is not a crypto total. And its aggregated per-coin charts are not real-time
by the vendor's own notice, while the markets table is. The same problem one layer down is
[where the bitcoin price actually comes from](https://cryptomarkets.tools/guides/where-the-bitcoin-price-comes-from).

**The history is shorter than the chart implies.** Binance's open-interest statistics endpoint says
plainly that only the data of the latest 1 month is available; the free funding-rate history is
deeper, and deeper still in the flat-file archive on the vendor's public file host. Bybit's open
interest reaches back to the symbol's launch and no further, which for a perpetual listed last
quarter is last quarter. Coinalyze's rolling window is the 25 to 33 hours described above. On
CoinGlass the ladder is the history: four-hour bars and coarser on the cheapest plan, and twelve
days is the deepest minute history sold off the shelf. Nothing here is a research archive, and
paying more mostly buys resolution rather than years.

**Pulling it and publishing it are separate questions.** The endpoints are open; the terms behind
them are not, and they reach the numbers rather than just the software —
[what exchange API terms actually let you do](https://cryptomarkets.tools/guides/exchange-api-terms) is the long version.
CoinGlass reserves commercial use for its third tier, and Velo's terms name derived data explicitly,
which covers a chart you publish from it.

## If you outgrow this

If the problem is **that you are polling**, buy a key rather than a bigger loop. CoinGlass sells the
cross-venue version of what you were assembling by hand, with a documented request ceiling and two
response headers that report your usage, and Velo has a usage-based tier at two cents a call that
does not appear on its pricing page unless the URL asks for it.

If the problem is **that the numbers must line up**, either take a product that has already done the
conversion — Velo's per-eight-hour normalisation is the clearest example in this catalogue — or own
it yourself, in writing: interval per symbol, unit per contract, venue list, and the date each was
last checked. The second is more work and it is the only version that survives somebody asking how
the figure was built.

If the problem is **depth**, that is an archive purchase rather than an API upgrade.
[Amberdata](https://cryptomarkets.tools/tools/amberdata) publishes a start date per dataset per venue, which is the useful
form: funding rates on 16 of its 20 futures venues with the earliest at May 2016 on BitMEX, open
interest on 21 rows, liquidations on only 10. [Coin Metrics](https://cryptomarkets.tools/tools/coin-metrics) carries funding,
predicted funding, open interest, liquidations and greeks across a 68-venue reference set, with a
Parquet-and-Snowflake delivery path beside the API.

If the problem is **options rather than perpetuals**, none of the above is the answer.
[Laevitas](https://cryptomarkets.tools/tools/laevitas) is the card with a calibrated volatility surface and greeks behind it,
and its key starts at 500 dollars a month.

And if what you actually want is **a screen rather than a feed**, the free web terminal at
[CoinGlass](https://cryptomarkets.tools/tools/coinglass-web) and the positioning metrics at
[Hyblock Capital](https://cryptomarkets.tools/tools/hyblock-capital) are the browser versions of this page's subject — Hyblock
gates its API to the Professional plan, at three months of history, or three years on the annual
plan. If candles are what you are really after,
[how to pull OHLCV from an exchange](https://cryptomarkets.tools/how-to/pull-ohlcv-from-an-exchange) is the neighbouring task.

## FAQ

### Do I need an API key for funding rates and open interest?

On the big perpetual venues, no. Binance's public market data carries the security type NONE, every endpoint under Bybit's /v5/market/ answers unauthenticated, and OKX documents its public data endpoints as needing no authentication. A key changes who is bound by which agreement, not what the funding and open-interest endpoints return. What is metered is your IP address rather than your credential.

### Why do two sites show a different open interest for the same coin?

Three reasons, and none of them is that one is wrong. They may cover different venue lists — sixteen at Coinalyze, thirty at CoinGlass, six at Velo. They may convert differently, because one exchange reports contracts, another the base coin and a third US dollars. And they may draw the line between stablecoin-margined and coin-margined contracts in different places, or not draw it at all.

### What is the difference between predicted and realised funding?

The predicted rate belongs to a settlement that has not happened yet and keeps moving until it does. OKX states that the fee is assessed with the funding rate calculated in the minute before assessment, so a rate read an hour earlier is a forecast of the payment rather than the payment. The realised series is what the funding-history endpoint returns, timestamped at settlement.

### How far back does free funding and open-interest history go?

Not far, with one exception. Binance's open-interest statistics say only the data of the latest 1 month is available; Coinalyze keeps 1,500 to 2,000 intraday datapoints and deletes the old ones daily. Bybit's open interest goes back to the symbol's launch. The exception is Binance's flat-file host, which carries futures funding-rate history as zipped CSV with no account at all.

## Sources

1. [Get Funding Rate History, USDⓈ-M Futures REST API](https://developers.binance.com/docs/derivatives/usds-margined-futures/market-data/rest-api/Get-Funding-Rate-History) — Binance, read 2026-09-21
2. [Open Interest Statistics, USDⓈ-M Futures REST API](https://developers.binance.com/docs/derivatives/usds-margined-futures/market-data/rest-api/Open-Interest-Statistics) — Binance, read 2026-09-21
3. [Open Interest Statistics, COIN-M Futures REST API](https://developers.binance.com/docs/derivatives/coin-margined-futures/market-data/rest-api/Open-Interest-Statistics) — Binance, read 2026-09-21
4. [Get Open Interest, V5 market endpoints](https://bybit-exchange.github.io/docs/v5/market/open-interest) — Bybit, read 2026-09-21
5. [Get Funding Rate History, V5 market endpoints](https://bybit-exchange.github.io/docs/v5/market/history-fund-rate) — Bybit, read 2026-09-21
6. [Get Instruments Info, V5 market endpoints](https://bybit-exchange.github.io/docs/v5/market/instrument) — Bybit, read 2026-09-21
7. [Introduction to perpetual swap funding fee](https://www.okx.com/en-us/help/iv-introduction-to-perpetual-swap-funding-fee) — OKX, read 2026-09-21
8. [Funding](https://hyperliquid.gitbook.io/hyperliquid-docs/trading/funding) — Hyperliquid, read 2026-09-21

*Last updated 2026-09-21. Corrected in place — an endpoint that moves is a bug here, not a new post.*