Which Hardware Wallets Support Liquid Staking Tokens

Every hardware wallet on this list can hold stETH, wstETH or rETH, because each is an ordinary ERC-20 token. The devices differ on what happens after that. Staking, unstaking and claiming are contract calls, and the useful question is whether the device decodes those calls on its own screen. On that measure the Ledger Nano X, at $149 in September 2026, is the cheapest documented route for a Lido user. It has a native Lido flow inside Ledger Live and a clear-signing rule strict enough to matter. The Trezor Safe 5, at $129, has decoded Lido calls only since 7 September 2026, and it stakes through a third-party dApp rather than through Trezor Suite.

The rest follows from that. The Ledger Flex adds nothing to liquid staking that its documents record, so it is $100 spent on something else. Rocket Pool appears in neither maker's published clear-signing list, so an rETH holder should treat its contract calls as unconfirmed on every device here. The Keystone 3 Pro cannot be assessed from the sources, for reasons set out below.

The Keystone 3 Pro has no column because none of the published sources used here document its staking flows, its clear-signing coverage or its price. A column of blanks would look like a finding. It would only show that the sources say nothing.

Ledger's strength is a rule. Its developer guide for Earn apps covers staking, liquid staking and restaking inside Ledger Live, and it states that "a Live App with even one un-clear-signed call will not be approved." Every stake, unstake, claim and delegate function must be declared with labels such as "Amount to stake", so that the Secure Screen shows text rather than hex. Lido is the first protocol named in the ERC-7730 registry for Earn. Swell is the one restaking-adjacent name on the list.

The staking flow itself is described in Ledger's Lido announcement. With the Ethereum app installed, the user opens Discover, selects Lido, enters any amount of ETH, reviews the fee and confirms on the device. The stETH that comes back appears in Ledger Live and can be swapped through Paraswap. Lido takes 10% of staking rewards, so stakers keep 90%. The yield behind that 90% is a separate subject from custody.

The weakness is the fallback. Outside the registry, a Ledger device blind-signs, and the user sees only raw hex. The wallet integration guide adds that EIP-712 messages, which are common in permit and delegation flows, blind-sign unless a matching eip712 descriptor exists. Separately, Crypto Valley Journal reports that Ledger had several customer-data breaches in 2025 and 2026. Those breaches touched customer records, not device keys, but a buyer may reasonably weigh them.

This device is for the Lido staker who wants to stake inside the companion app, with the least money spent on hardware.

For liquid staking, the Flex behaves exactly like the Nano X. Ledger's clear-signing documents describe Ledger devices and the ERC-7730 registry. They do not distinguish between models, so the Flex gets the same Lido flow, the same registry coverage and the same raw-hex fallback. The descriptor is fetched from the registry at signing time, with no app update needed, so a newly added protocol reaches both devices at once.

The Flex cost $249 in March 2026, which is $100 more than the Nano X. None of the sources gives that $100 a liquid-staking purpose. It suits a buyer who wants the Flex for reasons outside this question and holds LSTs as well.

Until September, a Safe 5 asked to sign a Lido transaction showed a hash or the words "Data Present", which left nothing to verify. Trezor's Clear Signing went live on 7 September 2026. It runs on the Safe 7, Safe 5, Safe 3 and Model T with Universal firmware 2.12.4 or later, and it works through Trezor Suite, WalletConnect and Trezor Connect without extra setup. Lido is on the launch list. So are Ethena and Lombard. Kiln and P2P are supported, but their ETH vaults are explicitly excluded from decoding. Trezor describes its own coverage as "registry-based, not universal".

Trezor Suite does not stake. Trezor's wstETH page lists send, receive and swap across Ethereum, Monad, Unichain and MegaETH. Minting stETH means connecting the Safe 5 to MetaMask or Rabby and using Lido's dApp. Lido's own help center describes the same split between holding and staking. Trezor's Rocket Pool pages confirm that Suite can send and receive rETH on desktop, web and mobile, and Rocket Pool lists Trezor as a supported device. Neither of those amounts to decoded Rocket Pool calls.

Trezor's guide names Lido as supported but does not list which Lido functions decode. Whether a withdrawal request and the later claim both render in plain language is not stated, and Ledger's lists are no more specific. This device is for a buyer who stakes through dApps anyway and wants the lowest price that still decodes Lido.

Little can be said about the Keystone 3 Pro from the documents behind this page. As an Ethereum hardware wallet that handles ERC-20 tokens, it falls under Lido's general statement: "Any hardware wallet that supports Ethereum can hold your stETH token." Beyond that, the sources do not establish whether it decodes ERC-7730 descriptors, what its screen shows for an uncovered vault, whether its companion software stakes natively, or what it costs. Until Keystone's own documentation answers those questions, nothing here supports recommending it for staking calls, and nothing supports ruling it out either.

Two things separate these devices, and neither is the hardware.

The first is where staking happens. Ledger puts Lido inside its own app and gates that app on every call being clear-signed. Trezor leaves staking to Lido's website and decodes what arrives. For a Lido user both routes now end in readable text, but only Ledger's has been policed by the maker for longer than a month.

The second is what happens at the edge of the registry, and that edge is where restaking vaults and newer LSTs sit. Both devices fall back to signing what they cannot read. Trezor at least flags it with a warning; Ledger shows hex. The difference between a blind and a clear signature matters most for the off-chain permit. Bex.co, citing Scam Sniffer, reports that Permit approvals accounted for 38% of 2025 losses above $1 million, and that the year's largest single theft, $6.5 million, used one malicious Permit signature.

The strongest objection is that this gap is temporary. Ledger, Trezor and MetaMask all read the same public ERC-7730 registry. Once a protocol publishes a descriptor, every device that supports the standard decodes it, so the coverage gap closes by itself. That is largely right, and it is the best argument for buying the cheapest device. It does not settle matters today, for two reasons. Trezor's list already departs from the shared registry (Kiln and P2P vaults are carved out). And a dApp screen can be rewritten by malware, which is how $1.5 billion left Bybit in February 2025. Convergence helps whichever device decodes the call at the moment of signing. It does nothing for a device that blind-signs it, and nothing for an approval the user has already granted to a vault.

A Lido staker who wants to stake from the wallet's own app should buy the Ledger Nano X. The Flex does the same job for $100 more. A Lido staker who already works through MetaMask or Rabby should buy the Trezor Safe 5 and update it to firmware 2.12.4 before signing anything. It is cheaper, and it now decodes Lido. An rETH or restaking-token holder gains nothing from either brand on contract calls, because neither maker names Rocket Pool or the major restaking vaults. Every device here holds those tokens safely. Any deposit, withdrawal or claim against them should be treated as a blind signature until the protocol's descriptor appears in the ERC-7730 registry.