KETJU Research

← The Register

staking

Benqi Staked AVAX (sAVAX)

Rejected
Max sleeve
Reviewed
2026-08-14 · v1
Next review
2026-11-15
Research basis
Individual research
Chains
Avalanche · crypto-backed
Symbols
SAVAX

Rejected venues wait the longest for re-review; a rejection has to earn another look before the scheduled date.

The protocol the Avalanche chain approval unblocked, reviewed the same day the chain cleared, and rejected on its own facts. sAVAX is Avalanche’s dominant liquid staking token ($146M at the 2026-08-14 survey), an exchange-rate token with a 10% reward fee, live since February 2022 with no exploit any source could find. Avalanche has no slashing, so downtime costs rewards, not principal. That is the whole good news, and none of it survives the exit test. Total sAVAX liquidity across every DEX is roughly $300k, about 0.2% of supply (largest single pool $171k, measured 2026-08-13), so the only real exit is the unstaking cooldown: 15 days, confirmed on-chain, before a client sees AVAX. The question a lending memo asks in a hurry, could your client get out today, has the answer no by design. History quantifies the cost: sAVAX traded at roughly a 4% discount in Terra week and 3% in FTX week, into depth that has since gotten thinner. Control fails independently. The sAVAX contract is an upgradeable proxy behind a 4-of-6 multisig with NO timelock and unpublished signers, read on-chain at this review; the staked AVAX itself moves through MPC infrastructure the docs do not identify; and the newest audit of the core staking contracts dates to 2022 (Halborn, Certora), four years and many releases ago, behind a $500k bounty guarding $146M. Validator distribution, the axis that decided the StakeWise review, is published nowhere: the docs describe a 65% random and 35% QI-voted split and no source reports the actual validator count or concentration. The system-level finding is the sharpest: 79% of all sAVAX sits as loop collateral, 43% in Benqi’s own lending market (itself rejected on the size floor) and 36% in Aave v3 Avalanche, at LTVs up to 95%. A stress deleverage sells into that $300k of depth. This is a well-run product whose position across Avalanche lending makes it a crowded exit with a two-week door, and no alternative exists to select instead: the one competitor, GoGoPool, shuts down 2026-08-15. A single-provider category with these controls is a category we decline.

The research file

The mechanism

sAVAX is an exchange-rate token, not a rebasing token: the docs say it automatically accrues rewards based on the staking rate, priced as total AVAX staked over total sAVAX minted, so a holder’s balance stays fixed while the AVAX claim per token rises. The rate was about 1.279 AVAX per sAVAX at 2026-08-13 (DefiLlama prices: sAVAX $8.27, AVAX $6.47), consistent with roughly four and a half years of compounding at 5 to 7 percent net. The protocol keeps 10 percent of staking rewards; there is no deposit or withdrawal fee.

The exit is a cooldown: 15 days of unstaking, then a 2-day claim window, after which unclaimed AVAX is restaked, per the docs. The 15 days is verified on-chain: the sAVAX contract’s cooldownPeriod reads 1,296,000 seconds, exactly 15 days, from contract 0x2b2C81e08f1Af8835a78Bb2A90AE924ACE0eA4bE on 2026-08-13. The 2-day claim window is docs-only; the matching storage variable was not located. Rewards keep accruing during the cooldown and stop during the claim window. The only instant exit is a DEX sale at whatever discount the pools bear; the docs describe no protocol instant-swap facility.

Avalanche has no slashing by design. A validator that misbehaves or goes offline forfeits rewards; delegated principal is never destroyed, and a validator must keep roughly 80 percent uptime to earn rewards at all, per Figment’s protocol page, with Kraken and Avascan staking guides agreeing. The slashing risk present on Ethereum is replaced here by lost yield from validator downtime and by the protocol layer: the principal-loss paths for sAVAX are contract compromise, key compromise, and MPC failure, not validator misbehavior.

Who controls it

The control facts in this section were read directly from Avalanche C-Chain RPC on 2026-08-13. sAVAX (0x2b2C81e08f1Af8835a78Bb2A90AE924ACE0eA4bE) is a TransparentUpgradeableProxy with implementation 0xb791c7a42fd0d10f90deaa906a8735f79719fa53. The proxy’s admin is a ProxyAdmin contract (0x2295e1cad2ea081a4a2ed85f59006e6fd42b5a66) owned by a Gnosis Safe v1.3.0 at 0x832c971f7c9f9a28789a934502010e74d8e489cd with 6 owners and a threshold of 4. A 4-of-6 multisig can upgrade the contract that holds the claim on all staked AVAX; the Safe owns the ProxyAdmin directly with no timelock in the upgrade path; and the 6 signer identities are published nowhere this review found. Exponential.fi’s protocol review reports the same shape: at least 4 signers, no timelock or none documented.

The docs’ risk page discloses the ability to pause markets, pause staking or redemptions, and adjust parameters on an accelerated timeline. Whether a pause can block a cooldown redemption already in flight is not documented. Movement of the staked AVAX between C-Chain and P-Chain runs through off-chain multi-party computation infrastructure; the docs name threshold compromise and operational halts as the risks, and publish neither the MPC operators nor the threshold.

Benqi was founded through Rome Blockchain Labs Inc by JD Gagnon, Hannu Kuusi, and Alexander Szul; the protocol launched April 2021 and liquid staking February 2022, per IQ.wiki, with Crowdfund Insider confirming Rome Blockchain Labs as the firm behind Benqi. Whether its personnel are the Safe signers is unknown. sAVAX itself needs no price oracle to accrue, but its use as collateral does: Benqi commissioned a Chaos Labs dual-oracle contract audit in May 2025, and third-party reporting describes two independent price sources with audited discrepancy handling. The exchange-rate reporting path from P-Chain to the C-Chain contract is not documented.

Validator delegation

The docs split delegations 65/35. Sixty-five percent goes to an Open Pool allocated randomly across validators meeting three criteria: at least 6,000 AVAX staked, high-spec hardware, and at least 15 days of uptime, at a 2 percent delegation fee. Thirty-five percent is the BENQI Miles Pool, formerly veQI, where QI stakers vote AVAX toward specific validators needing at least 2,000 AVAX, 15 days of uptime, and a delegation fee between 2 and 5 percent. The docs specify no cap per validator.

Who chooses: for the 65 percent, an algorithm over an eligibility set Benqi defines; for the 35 percent, QI token voters. In both cases Benqi’s contracts and MPC infrastructure execute the delegation. The number of validators actually receiving delegation, and the concentration among them, is published nowhere this review found: not in the docs, not on a dashboard. This is the same hole the StakeWise review treated as decisive, and it is open here.

The record

Live since February 2022; DefiLlama’s first TVL point is 2022-02-17 at $57.6M. Monthly samples: $87M May 2022, $53M June 2022, $108M January 2023, $261M January 2024, $305M January 2025, $251M January 2026, and $146.0M at 2026-08-13. TVL has halved in 2026, largely with the AVAX price. No exploit of Benqi liquid staking or lending was found in any incident database or news search; third parties state it plainly, no major protocol exploit since launch in August 2021, per a dev.to security writeup with Exponential.fi concurring. That is absence of evidence after a deliberate search, not proof.

The audit list, per the docs’ risks page: Halborn on lending in May 2021, Halborn on liquid staking from November 2021 to February 2022, Certora formal verification of sAVAX in April 2022, Certora again for the Aave v3 listing in June 2022, and Chaos Labs on the dual-oracle contract in May 2025. The core liquid staking contracts have had no fresh audit since 2022 as far as the docs disclose. The Immunefi bounty maxes at $500,000 with 38 assets in scope, program page last updated 2026-08-05; whether the sAVAX contract specifically is in scope was not confirmed line by line.

The exit

Secondary liquidity is thin to the point of being decorative. GeckoTerminal on 2026-08-13 showed the largest sAVAX pool at about $171k of reserve, the next at $73k, and roughly $300k total across all sAVAX pairs on Avalanche. Against $146M of TVL, on-chain exit liquidity is about 0.2 percent of supply, and a client-sized sale of $100k and up would move these pools several percent. The DEX instant-exit path is effectively closed; the 15-day cooldown is the exit. The contract-level exit is always 1:1 at the exchange rate after 15 days plus the 2-day claim window, subject to the pause powers above.

The commonly cited 2022 depeg is real but modest, and no named write-up of it was found. From DefiLlama historical prices: on 2022-05-12, Terra collapse week, sAVAX traded at 0.973 AVAX against an exchange rate near 1.01, a discount of roughly 4 percent; on 2022-11-10, FTX week, the ratio was 1.005 against an exchange rate near 1.04, a discount of roughly 3 percent. No deeper episode surfaced in price data or reporting. Context makes the Terra number meaningful: sAVAX had been accepted as collateral on Terra’s Anchor in February 2022 (TerraSpaces; Benqi’s Medium), so the Terra unwind was a direct stress test, and the discount stayed in single digits.

Collateral use and circularity

On-chain balances read 2026-08-13: total supply 17,710,077 sAVAX, of which 7,617,382 (43.0 percent) sits in Benqi’s own qisAVAX lending market (0xF362feA9659cf036792c9cb02f8ff8198E21B4cB) and 6,406,562 (36.2 percent) in Aave v3 Avalanche’s aAvaSAVAX (0x513c7E3a9c69cA3e22550eF58AC1C0088e918FFf). Seventy-nine percent of all sAVAX is collateral in two lending markets, which is why the DEX float is thin.

The dominant use is the sAVAX-collateral, AVAX-borrow loop. Aave’s Avalanche e-mode runs sAVAX at 92.5 percent LTV and a 95 percent liquidation threshold, per the Aave governance forum, and third-party reporting (team1.blog) has Aave v4 on Avalanche listing sAVAX at a 95 percent collateral factor borrowing wAVAX only. Benqi Lending itself holds $83.9M of TVL per DefiLlama. The circularity is direct: the largest holder of sAVAX is Benqi’s own lending market, a venue rejected on the size floor, and the token’s demand and its liquidation flows both route through leverage on the same two protocols. A forced deleverage of the loop sells sAVAX into the roughly $300k of DEX depth.

The comparison

There is no comparison left to run. GoGoPool, the second Avalanche liquid staking protocol, rebranded to Hypha and shuts down on 2026-08-15; its end-of-life page tells users to withdraw minipools and redeem stAVAX for AVAX, and its remaining TVL is $3.6M (DefiLlama, 2026-08-13). Benqi at $146M is the only at-scale AVAX LST; in June 2024 the two together held nearly $300M, per an official Avalanche post.

This cuts both ways: there is no better-distributed alternative to prefer, and no fallback if Benqi trips a kill criterion. Approving sAVAX means approving a category with a single provider, run by one company (Rome Blockchain Labs), upgradeable by one 4-of-6 Safe with no timelock. For fee scale: Benqi keeps 10 percent of rewards, Lido on Ethereum keeps 10 percent, and Rocket Pool node commission averages around 14 percent.

Open questions

How many validators currently receive Benqi delegation, and at what concentration. No dashboard or disclosure exists, and this was the decisive axis in the StakeWise review. Who the 6 signers of Safe 0x832c…89cd are, and whether the same Safe controls the lending market and the MPC infrastructure. Whether the pause power can block a cooldown redemption already in flight; the docs disclose pause of staking or redemptions without detail. The MPC threshold and operator set for C-Chain to P-Chain transfers of the staked AVAX.

Whether any part of the liquid staking system has been re-audited since the 2022 Halborn and Certora work; the docs list nothing newer for these contracts except the 2025 oracle audit. Whether the sAVAX contract itself is in the Immunefi scope, and whether $500k is a sane bounty against $146M at stake. The 2-day claim window needs an on-chain read, including what exactly happens to a claim that lapses; the docs say restaked, unconfirmed in code. And where the DEX liquidity went: Trader Joe and Pangolin once carried deep sAVAX pools, and today’s $300k needs an explanation, whether incentive withdrawal, migration, or concentrated liquidity this review mismeasured.

Sources

The claims above trace to these. Where a number could not be independently verified, the thesis says so.

Inherited controls

The verdict above grades the protocol layer. Every position also inherits the asset it holds and the chain it settles on. The least safe layer sets the position’s grade, and the position table names which one that is.

ChainVerdictGradeControl constraint
AvalancheApproved · limits crypto-backed no party can freeze or seize C-Chain funds, but one vendor writes the only production client and Messari measured over a third of stake hosted on AWS.
The memo is public. The watching is the product: the terminal reads your clients’ wallets against this Register and flags the events above when they fire. $49 per advisor per month, first 14 days free. Start the trial.