Reading Gate.io Against Its Own Public Data: What a 20% Fee Discount Is Actually Worth

Gate.io is a venue you can judge from its own public data, and that data says two things at once: its spot book in the majors is genuinely deep, and its long tail of listings is genuinely empty. This article contains the publisher’s referral link, and if you use it the publisher earns a commission. That does not change the numbers below; it is why you should read them like an auditor.

Every figure here was measured between 07:16 and 07:50 UTC on 2026-10-05, from Gate’s own public endpoints and from CoinGecko’s exchange record. Where a claim comes from marketing rather than a machine-readable API, I attribute it to the company. Where I could not verify something, I say so and show you how to check it.

What Gate.io is, and what it has been since 2013

Gate has been operating since 2013, and that is the most useful single fact about it. I have seen many exchanges announce themselves in a good year and vanish in a bad one. A venue that has kept the lights on through several market cycles has already answered a question no launch can. Longevity is the fact. Size is the noise.

CoinGecko’s exchange record for Gate lists year_established 2013, country Panama, trust_score 10, trust_score_rank 5, centralized true, has_trading_incentive false, with 1,523 coins and 1,670 pairs. Those counts are CoinGecko’s tally of the markets it tracks, not Gate’s advertised listing number, and the two should not be confused. The record is on CoinGecko’s public Gate record.

On raw BTC volume at the same moment, Gate did 15,238 BTC over 24 hours. Sorted that way across the top 60 exchanges CoinGecko returns, it sits 7th. In the same table Binance did 81,824 and took 1st, OKX 20,610 and 3rd, Bybit 14,956 and 8th, and Coinbase Exchange 13,307 and 11th.

Be exact about what that ranking is. A trust score is a heuristic built from traffic and reported order-book data. A volume rank is a position in a table on one afternoon. Neither answers solvency, nor whether your withdrawal clears next Tuesday.

The three gates: a trading screen is not a machine

Every exchange presents the same face, and the face is the screen. I have watched people choose a venue by how the order form feels, which is like choosing an airline by the colour of the boarding pass.

Years ago I wrote that software development has three gates. Gate one is business function: the thing does what it says on the box, and almost anyone who can code gets through it. Gate two is business performance, where the craft lives; passing it takes operating systems, networking, data structures and algorithms. Gate three is business intelligence, where the system tells you things you did not already know.

For an exchange, the trading screen is gate one, the part everyone judges. The order-matching engine, the risk engine, the funding settlement and the withdrawal pipeline are gate two. Nobody photographs them, and they are where a venue succeeds or fails, because they must hold under load, under attack, and under a hundred thousand users refreshing at once.

This is where benefit-first thinking applies. Judge any architecture or product choice by the return it produces: a faster release, a better service level, a lower human cost. Never by how elegant it looks. I have seen beautiful systems that needed more people than the systems they replaced, and that is a loss no diagram can hide.

The long tail is the product, and the order book is the price

Gate’s profile records over 5,200 cryptocurrencies, and the company reports more than 12,800 stock assets. A listing count is a marketing number. An order book is a physical one, and the physical one is what you trade against.

I pulled the top 100 levels of the public spot order book for nine pairs at one moment and summed the notional within 1% of the best quote. The shape is the argument.

Look at the bottom row. RAIN_USDT traded about 39 million tokens of 24h base volume, which sounds busy, and yet only about USD 188 of asks sat within 1% of its best quote. Its 24h range was USD 0.010551 to 0.013599, roughly a 29% swing on that book. Volume and liquidity are two different measurements, and a long tail of listings makes exactly this shape.

The venue’s own public spot pair metadata says the same in parameters. On BTC_USDT the recorded slippage is 0.03 and the market_order_max_money is 5,000,000; on RAIN_USDT they are 0.1 and 500,000. The daily band, up_rate and down_rate, is 0.08 against 0.4, or 8% against 40%. A market order in the tail is both more constrained and permitted to move five times as far in a day.

What you buy with a long-tail venue is the listing count. What you pay, if you trade the tail, is the thin book. I took that trade-off apart in our earlier piece on what a long-tail listings exchange costs you. Here I only want the number.

What a 20% fee discount actually buys you

A 20% discount on trading fees is real, and next to the cost most retail traders never measure it is small. Round trip means open plus close, so two sides. Using the measured futures taker rate of 0.075% per side:

So a 20% discount on futures fees is worth USD 3 per USD 10,000 round-tripped as a taker. Hold that number. On RAIN_USDT the measured spread was 0.6331%. Crossing it once per side on a USD 10,000 notional costs roughly USD 99 — about 33 times the USD 3 fee saving.

On BTC_USDT the measured spread is 0.0003%, so the same round trip pays only cents in spread, and the discount is worth more than the entire spread cost. The discount is worth having where the book is deep. It is noise where the book is thin.

This is benefit-first reasoning applied to a fee. A discount is a real benefit only when it is larger than the cost you are not looking at. Most traders stare at the fee and never at the spread, which is why the fee is the thing that gets advertised.

Reading the fee schedule before you deposit

The instruction takes about ten minutes. Do it before you deposit anything, because a fee claim you cannot see in your own account is a slogan. Start with Gate’s public futures contract list; it needs no key and no account.

That endpoint returns 1,028 contracts, and across all of them there is exactly one distinct value for each of four fields. maker_fee_rate is -0.0001, minus one basis point, meaning the venue pays liquidity providers. taker_fee_rate is 0.00075, or 0.075%. ref_rebate_rate is 0.2. ref_discount_rate is 0.

Now the honest part. ref_rebate_rate = 0.2 is a field in Gate’s own public metadata, and it is exactly the number a referral page turns into 20%. But in referral programmes a 20% figure can mean one of two opposite things: a discount off the referee’s fees, or a commission paid to the referrer. Both are advertised as 20%, and ref_discount_rate is 0 in that same payload. The publisher’s link is represented as 20% off spot and futures fees. I cannot verify from public metadata that it is a referee discount, and Gate’s marketing fee pages returned HTTP 403 to automated fetching. So I will not call a spot base fee schedule measured, and I will not call the 20% verified. The one unambiguous place is your own account’s fee table, after applying the link, before you deposit.

1. Open https://api.gateio.ws/api/v4/futures/usdt/contracts in a browser tab and confirm it returns 1,028 contracts.
2. Read maker_fee_rate and taker_fee_rate off any single contract in that payload.
3. Search the same payload for ref_rebate_rate and ref_discount_rate, and note that ref_discount_rate is 0.
4. Create the account and apply the referral link at https://www.gate.com/share/zmkmsvip before any deposit.
5. Open your own account fee table for one spot pair and one futures contract, and write down the maker and taker numbers you actually see.
6. Compare the two sets. If taker_fee_rate did not move from 0.075% after applying the link, the discount did not apply to you.
7. Check funding_interval for the contract you intend to trade; it is 28800 seconds, every 8 hours, on every contract measured.
8. Read the per-pair parameters for any tail listing you plan to trade: slippage, market_order_max_money, up_rate and down_rate.

That is the whole method. Your account fee table is the only document that matters, and it is the one a marketing page cannot rewrite.

Funding, tiers and the switches a discount does not touch

The futures contract list tells you things a fee page will not. Every one of the 1,028 contracts measured carries funding_interval = 28800 seconds, every 8 hours, without exception. Across those contracts the current funding rate takes 158 distinct values, including zero, small negatives and small positives. The interest_rate field takes two values, 0 and 0.0003.

Funding is not a fee, and a referral discount does not touch it. On a USD 10,000 position, a funding rate of 0.01% per interval is USD 1 per interval, about USD 3 per day. That 0.01% is illustrative, not measured; the measured set is those 158 values, and funding moves between longs and shorts.

The maker side is the interesting one. maker_fee_rate is -0.0001 on every contract, minus one basis point, so liquidity providers are paid rather than charged. Trade with limit orders and you sit on the other side of the table from a taker, which is where the venue’s economics want you.

Leverage is per contract and not uniform. BTC_USDT and ETH_USDT go up to 200x, SOL_USDT and XRP_USDT to 100x, SHIB_USDT to 75x, while GT_USDT is only 5x, and the minimum is 1x everywhere. A venue that sets its own token at a fraction of the leverage it offers on Bitcoin is telling you how it models risk.

Then come the switches a discount sits underneath rather than beside. Gate’s own SDK documentation for the authenticated spot fee endpoint lists maker_fee, taker_fee, gt_maker_fee and gt_taker_fee, where the gt_ fields are the rates when GT deduction is enabled and are documented to return 0 when it is off. VIP tiers sit above the base schedule. Whether a 20% discount stacks on a VIP rate and a GT deduction is what your own fee table answers and a marketing page does not.

Paraphrasing a line he repeated often: if an architecture needs more people to keep it alive than the system it replaced, it has failed, however elegant it looks.

Hao Chen, CoolShell, paraphrased

The point of that blockquote is not nostalgia. Count the moving parts a venue accepts as normal, then ask what each costs you in withdrawal latency and in the price you cross.

Custody, licences and the claim you actually hold

An exchange balance is not a coin. It is a claim against a company, and a company is a legal person that can fail. I have seen many firms that looked permanent on a screenshot and were not permanent in a court. When a venue lists its licences, read them as claims and ask which entity each one attaches to.

As recorded by CoinGecko, Gate lists registrations, licences, authorisations or approvals in Malta, Italy, Gibraltar, the Bahamas, Hong Kong, Japan and the UAE. In 2024, says the record, Gate Group completed the acquisition of a Japan-licensed exchange. In April 2025 Gate Technology FZE obtained a full operational licence from Dubai’s VARA. In September 2025 Gate Technology Ltd was authorised under the EU MiCA framework by the Malta Financial Services Authority. In February 2026 Gate’s Malta entity obtained a Payment Institution licence under EU PSD2 from the MFSA. Those are the record’s claims, in the record’s words, attributed rather than asserted.

Proof of reserves has the same shape. Gate states that it uses zero-knowledge technology to provide verifiable proof of reserves. As of 2026-06-22 it reported total reserves of approximately USD 8.182 billion, a reserve ratio of 115%, covering nearly 500 types of user assets. Read every clause: stated by the company, reported as of a date. A ratio is a snapshot. Custody is a continuing obligation.

Data beats experience; experience beats intuition.

On reading a venue’s own public metadata instead of its marketing page

Three questions I name rather than answer. Whether the 115% ratio is verifiable by a customer for a specific asset at a specific time, and who signed it. Whether a 20% discount stacks on top of a VIP tier and a GT deduction, and what the effective floor is. And the one that matters most: which entity in the Panama, Malta and Dubai group actually holds a customer’s assets, and where that customer’s claim ranks in line. A logo answers none of these.

The standard I would apply

What is this venue genuinely good for? I think it is good for a trader who works the deep end of the book, BTC and ETH and the few pairs with six-figure depth inside 1%, and who wants a maker rebate, an eight-hour funding schedule, and the oldest operating date on the public record. It is not good for someone who trades the long tail and expects a fee discount to matter there, because there the spread eats the saving many times over.

The three gates give the whole standard. Anyone can pass gate one, and Gate has. Passing gate two for a decade is expensive and rare, and that is the part I take seriously. Gate three, whether the platform tells you something about your own trading you did not already know, is where you watch what it does with your data, and I cannot judge that from a fee table.

So the one standard I apply to every venue, this one included: do not trust a ranking, a logo or a slogan. Open the endpoint, read ref_rebate_rate and ref_discount_rate with your own eyes, apply the Gate.io referral link if you choose to, then read your own fee table and your own order book before you deposit a single unit. The link is here because the publisher is paid when you use it, and that is the whole of the arrangement.

None of this is investment advice. I have not told you whether to buy, sell or hold anything, including GT, and I will not.