Crypto interest rates and the conditions attached to them

Crypto interest rates are advertised the way airline fares used to be: a striking number at the top, and the conditions that determine whether you can have it somewhere further down. This page collects the rates we were able to verify from provider documentation on 16 September 2026, and puts the condition attached to each one in the same row. It does not explain how staking or lending work — those live oncrypto staking and crypto lending. This is the arithmetic page.

Start with the benchmark, because without one no rate means anything. If a US dollar earns 3.97% for three months with no credit risk, then a stablecoin paying 3.5% in a smart contract is not a yield opportunity. It is a discount you are paying for on-chain settlement.

The benchmark ladder

Every rate on the rest of this page should be read against this table. These are the returns available on short-dated US government debt, plus the aggregate yield of the tokenised version of the same thing.

The last row is the interesting one. The tokenised treasury market held roughly$15.65bn and paid an aggregate 3.74%, which is below the 3.97% on the underlying bills. The wrapper costs somewhere between 25 and 50 basis points. What you buy with that difference is round-the-clock settlement and the ability to use the position inside DeFi — not extra return. Tokenised treasuries goes through the individual funds.

APR, APY and why the difference is not pedantry

APR is the simple annual rate: a rate applied once, with no assumption that you reinvest anything. APY is what you finish the year with after earned interest starts earning interest of its own. The formula is APY = (1 + APR ÷ n) to the power of n, minus 1, where n is the number of compounding periods a year.

A 6% APR paid and redeposited daily produces a 6.18% APY. Paid weekly, 6.18%. Paid monthly, 6.17%. At 6% the difference is about 18 basis points and rarely decides anything. At 12% it becomes 12.75%, and at the promotional rates some venues advertise the divergence is enormous. The practical questions are therefore: how often is interest credited, and does it go back into the earning balance automatically?

Answers differ more than you would expect. CEX.IO publishes its savings formula openly as reward rate × amount ÷ 365, credited daily, and redeeming on a given day forfeits that day's reward. Bitpanda credits staking rewards weekly and auto-restakes them, which compounds. Kraken accrues daily and pays weekly. Bybit accrues hourly, distributes daily at 00:30 UTC, and offers an auto-savings option that compounds. Nexo's flexible product compounds daily; its fixed-term product pays a single lump at maturity and compounds only if you auto-renew. Ledn accrues daily and pays monthly, and its own documentation does not address compounding at all. A rate quoted as APY with no stated compounding frequency is doing some work for the marketing department.

The four conditions that shrink a headline rate

1. The balance tier

This is the most common and the most effective. Bybit's flexible USDT product advertises12.00% APR. That tier ends at 500 USDT. From 500 to 1,000 USDT the rate is 0.70%, and above 1,000 USDT it is 0.28% — so essentially all of any serious balance earns 0.28%, not 12%.

Crypto.com uses the same idea with a smoother gradient. For fixed-term products on every asset except USDC, the full advertised rate applies to the first US$3,000; the nextUS$27,000 earns half that; and anything above US$30,000 earns 30% of the reduced rate — which is 15% of the headline. Run that across a $100,000 allocation and the blended multiplier is 0.27: on our arithmetic from Crypto.com's own published tiers, you receive about 27% of the advertised rate on the whole balance. Independent rate tracking on 23 August 2026 put Crypto.com's realised base rates at 0.5% on USDC and USDT, 0.2% on BTC and ETH and 0.25% on SOL, against headline figures of up to 1.5% on BTC and up to 4% on ETH.

2. The commission

For staking products the advertised APY is frequently the gross network rate, with the platform's cut deducted afterwards. The range across this market is enormous — 0% to 90% — and it is the single largest determinant of what a staker receives. The full table is further down this page.

3. The proportion actually deployed

Kraken's flexible staking page states that "Kraken will only stake a portion of your assets. You will receive rewards on up to 50% of the assets you choose to stake." The advertised APY is truthful; it simply applies to half of what you committed. Nothing else on this page changes an effective yield by a factor of two in a single sentence.

4. Nominal versus real

Staking rewards on most proof-of-stake networks are newly minted tokens, so the correct calculation is (1 + nominal) ÷ (1 + inflation) − 1, not nominal minus inflation. Cosmos advertises about 19.64% against token issuance of 12.67%, which is a real yield of 6.19%. The naive subtraction gives 6.97% and overstates it by 78 basis points. Sui is the case that breaks the intuition entirely: 1.49% nominal against 2.54% issuance is a negative real yield of −1.02%, so the balance grows while the share of the network shrinks. The cross-asset version of this sits onstaking rewards compared.

CeFi rates: what the big custodial platforms advertise

This is the table the rest of the page exists to support. Every figure comes from the provider's own published materials on 16 September 2026, and the fourth column is the condition that determines whether you personally get it.

DeFi rates: the same asset, two venues

DeFi rates are not set by anyone. They are computed from utilisation — the share of a pool currently borrowed — which is why they move continuously and why the same asset can pay very differently in two protocols on the same chain at the same moment.

The bottom three rows are the most under-reported numbers in this entire subject. Bitcoin wrappers and staked-ETH wrappers supplied to Aave earn essentially nothing, on billions of dollars of deposits, because nobody borrows them. People borrow against them. Their utilisation rates — 0.5%, 2.3%, 0.7% — say it plainly. If a platform is advertising a meaningful BTC interest rate, the money is not coming from a lending market like this one.

The USDC row shows the other side of the problem: a range, not a point. Two reputable trackers gave Compound V3's Ethereum USDC market 4.61% and 5.76% at the same timestamp, with a seven-day range of 3.29% to 5.85%. That is not an error in either tracker. It is a rate that genuinely moves intraday with utilisation, and quoting it as a single figure would be the error. Elsewhere in DeFi the same variance appears at the protocol level: Morpho's 245 tracked pools averaged 4.28%, Maple's three averaged 4.89%, Spark's 22 averaged 2.01%, Kamino Lend's 151 averaged 1.93% and Euler's 151 averaged 9.98% — the last of those mostly long-tail collateral on Monad, with incentives included.

For a like-for-like ladder of yield-bearing stablecoins on the same date: OUSG 3.44%, USTB 3.59%, USDY and Sky's sUSDS both 3.60%, Aave USDC 3.57%, Steakhouse's curated USDC vaults 3.18% on Base and 4.36% on Ethereum, Ethena's sUSDe around 5.03%, Sky's stUSDS 5.11% and Sky Fixed Yield 4.85%. Each of those is a different risk. Stablecoin yieldexplains which is which.

Staking commissions: the fee table nobody publishes together

A staking commission is charged on rewards, not on principal, which makes it sound smaller than it is. On a gross Ethereum APR of about 2.55%, here is what each published rate actually costs per year.

Put the bottom of that table next to the benchmark at the top of the page and the problem is obvious. Ethereum's inflation-adjusted real yield is about 1.66%. A 35% commission takes roughly 89 basis points of it — more than half. A 90% commission takes 230 basis points, which is more than the entire gross rate would have been after inflation. The commission is not a detail attached to the rate. On a 2.5% asset, it frequently is the rate.

The disclosure picture is worse than the fee picture. A survey of 14 providers on 12 September 2026 found that only five published a specific fee number: Lido, Everstake and Allnodes at 10%, P2P.org at 5%, and stakefish at 0% on consensus rewards with 50% on MEV. Kraken, Bitpanda, Kiln, StakeWise, Rocket Pool, Figment, Coinbase and Bitvavo did not publish a percentage on the pages checked; Coinbase and Bitvavo returned HTTP 403 to the fetch. Coinbase and Kraken's numbers above come from support documentation rather than the marketing pages a prospective customer reads first.

When a provider's own pages disagree

This happens often enough to be a category of its own, and the resolution is usually mundane rather than sinister.

• Nexo advertises "up to 15%" on its loyalty page and "up to 16%" on its US page, while the highest flexible rate in its own published asset table is 13% on DOT. The three numbers are mutually inconsistent and Nexo does not publish a per-tier, per-asset table that would reconcile them.
• Gate shows USDT at 3.80% for a VIP3-14-gated seven-day term in its featured panel and 7.06% labelled "Event Bonus" in the market table immediately below it, on the same page, with a separate "Limited-Time 4% APR on USDT" promotion running alongside.
• Kraken shows ETH at 1.31% on its Auto Earn page and 2.63% for bonded staking on its staking-coins page. These are genuinely different products with different liquidity terms, so it is not a contradiction — but two ETH numbers on one website, a factor of two apart, is not a helpful way to present it.
• CEX.IO markets locked savings at 30, 60 and 90 days with higher rates on its savings page, while the FAQ on that same page states: "Presently, we don't provide a Locked Savings option." Only flexible savings was actually available when we checked.
• KuCoin displays USDT as a range of 0.45% to 100%. The top of the range is a promotional quota product, not a rate anyone receives on a normal balance.

Why the date stamp matters more here than anywhere else

Every number on this page has a shelf life, and the shelf lives differ by an order of magnitude. DeFi rates recompute every block. Binance states its flexible APR is "subject to change every minute" and its bonus tiered APR "can change daily". Coinbase says its USDC rate may change at any time, with changes typically effective on the first calendar day of the month, and that it can vary by region. Ondo resets USDY monthly. Sky's savings rate is set by governance vote, and its history proves the point better than any argument: 8% in August 2023, 6% around August 2024, 12.5% in December 2024, and 3.60% when we checked.

Rates also disappear. Crypto.com's stablecoin Earn now shows "up to 0%". Coinbase paywalled USDC rewards on 15 December 2025. Ledn discontinued BTC and ETH Growth Accounts on 1 July 2025. CEX.IO discontinued Automated Staking on 1 April 2025. A comparison table with no date on it is not a comparison table; it is an archive. Ours says 16 September 2026, and by the time you read this some of it will have moved.

If you want the products rather than the numbers, every earn platform we reviewed has the licences and exclusions alongside. If you want to know which of these rates is legal where you live, availability by country covers it, and regulation of crypto yieldexplains why several of the stablecoin rows above are under pressure. For deciding how much to put behind any of them, passive income with crypto is the sizing discussion, and crypto savings accounts covers the product category in full.