GnosisDAO is the community-run treasury and governance body of the Gnosis ecosystem — one of the oldest continuously-operating organizations on Ethereum. It began not as a DAO but as a prediction-market protocol: Gnosis was founded in 2015 by Martin Köppelmann and Stefan George inside the ConsenSys incubator, and its April 2017 ICO raised 250,000 ETH (~$12.5M at the time) in about twelve minutes — a landmark early token sale. In December 2020 the project handed control of its treasury and roadmap to GNO holders, converting a company-led protocol into a protocol DAO that has since become a holding entity for a portfolio of spun-out products.

Oracle-executed governance: SafeSnap and Zodiac

GnosisDAO's most-copied contribution is a governance pattern that keeps voting cheap while making execution trustless. Proposals (Gnosis Improvement Proposals, "GIPs") are drafted on the forum, then ratified by a gas-free Snapshot vote — off-chain signalling that costs holders nothing. The innovation is what happens next: SafeSnap — now the Zodiac Reality Module — attaches the executable transactions to that Snapshot vote and posts the outcome to the Reality.eth escalation-game oracle. Once the oracle finalizes the result and a cooldown passes, anyone can trigger the on-chain transactions permissionlessly from the DAO's Safe. This "optimistic governance" bridges the gap between free off-chain voting and binding on-chain action without a trusted multisig deciding what the vote "meant" — GIP-11 enabled it for GnosisDAO, and it now underpins dozens of DAOs. Compare it to the fully on-chain and delegate-based models in DAO governance models.

The ecosystem it spun out

GnosisDAO functions less like a single product and more like an on-chain holding company: tools built for internal use were repeatedly graduated into standalone projects, several with their own tokens and DAOs.

• Safe (formerly Gnosis Safe) — the dominant smart-contract wallet / multisig standard, securing $100B+ in assets; spun out with its own SafeDAO and SAFE token.
• CoW Protocol — batch-auction, MEV-protecting DEX aggregator (CoW Swap), now governed by CoW DAO and the COW token.
• Gnosis Chain — an EVM Layer-1 (the former xDai chain) where GNO is the staking asset for validators.
• Gnosis Pay — a self-custodial Visa debit card settling directly from an on-chain Safe.
• Circles — a personal-currency / universal-basic-income money system.
• Zodiac — the open standard behind SafeSnap and a toolkit of composable DAO modules used far beyond Gnosis.

The redemption votes: GIP-150 and GIP-151 (2026)

GNO has a fixed 3,000,000 supply and a treasury that has long been worth more per token than GNO traded for, which made GnosisDAO the case study for a question every token-treasury DAO eventually faces: does the token grant a claim on the assets behind it? In 2026 the DAO answered it twice, seven weeks apart, and got opposite results.

GIP-150 was rejected. Proposed in April 2026 by a group of activist holders the press dubbed the “RFV Raiders”, GIP-150 offered a voluntary, one-time redemption of GNO for a pro-rata share of the treasury – liquid buckets in kind, plus a synthetic gLTD-CLAIM token standing in for future distributions from the illiquid venture book and Gnosis Ltd’s enterprise value. It lost on Snapshot with 167,692 GNO for, 432,545 against and 1,611 abstaining, on 95 votes.

GIP-151 passed by dropping the contested half. Filed in June 2026 and this time “developed jointly with the Gnosis founding team”, it kept the one-time pro-rata redemption and deleted the synthetic claim token along with any attempt to value the operating business – a simplification the proposal justifies in its own words as avoiding “any dispute over the valuation of illiquid or strategic holdings”. It carried on Snapshot with 157,749 GNO for, 2,500 against and 1,492 abstaining on 49 votes – 161,740 GNO in total against a 75,000 GNO quorum, or about 216% of it.

What is actually redeemable. The redemption is not against the headline treasury. GIP-151’s own worked example puts the treasury excluding GNO at $151,220,464 and values the GnosisVC portfolio at $5,200,000 – a 60% discount to the ~$13M called across two of three capital calls – for a redeemable treasury of $156,420,464. The eligible base is 1,364,058 GNO: the 3,000,000 total supply less the DAO’s own holdings and Gnosis Ltd’s, both excluded from the numerator and the denominator. That works out to a net asset value of roughly $114.67 per eligible GNO. Redeemers take ETH and stablecoins for the liquid and long-tail buckets, SAFE, COW and HOPR in kind, and stablecoins for the GnosisVC slice; redeemed GNO returns to the DAO permanently, so every holder who stays ends up with a larger claim on what remains.

A redemption interface went live alongside the Noca treasury dashboard that supplies the NAV reference. The pair of votes is the industry’s clearest live test of exit rights at treasury scale, and the arithmetic of the two tallies is the part worth keeping: the winning vote drew about a quarter of the weight and half the voters of the one that defeated it, and the opposition fell from 432,545 GNO to 2,500. Retrofitting an exit right onto an established DAO turned out to be less a fight over whether members may leave than over what they may carry out with them.

The 2026 docket: a spin-out and a rollup

Redemption was not the only structural question on GnosisDAO’s 2026 ballot. Two proposals since have asked the DAO to change what it owns and what its chain is.

GIP-152 – spinning out the Gnosis App. Passed on 12 August 2026 with 105,103 GNO for, 3,353 against and 51 abstaining on 49 votes, GIP-152 moves the consumer app out of Gnosis Ltd’s annual funding mandate and into an independent company run by the team that built it. GnosisDAO invests $1.5M in cash and contributes the product and IP at a $1M valuation, taking a $2.5M SAFE position at a $10M post-money cap. The proposal is candid about why: the app “hasn’t yet found breakout success/PMF”, and with the GIP-128 renewal approaching the choice was to shut it down or restructure it. It is the holding-company pattern applied in reverse – converting an internal cost line into a priced equity position.

GIP-153 – Gnosis Chain as an Ethereum Economic Zone rollup. GIP-153, authored by Friederike Ernst with Philippe Schommers and Ben Carvill, asks the DAO to align on transitioning Gnosis Chain from a standalone Layer 1 into a ZK-proven rollup settling natively on Ethereum, ending the treasury-funded staking subsidy in favour of fee capture and targeting genesis around the turn of 2026/27. Its motivation section is unusually blunt for a governance document: “Gnosis Chain has failed to deliver on its original value proposition”, because credible neutrality “is Ethereum’s home turf”. No funds are requested; the vote seeks direction only. It closed on 19 August 2026 at 12:55 UTC with 123,424.71 GNO on 54 votes – 123,158.45 for, 115.27 against, 150.99 abstaining – clearing the 75,000 GNO quorum at 165% (final scores read from Snapshot's GraphQL API on 3 September 2026, proposal 0x5522f4cb…). The day before, it looked nothing like that. Read off Snapshot on 18 August it stood at 7,659.53 GNO for and 115.27 against on 34 votes – 7,774.80 GNO in all, 10.4% of the bar – with 94% of that weight in three addresses and the largest alone carrying 5,000.48 GNO, 64% of everything then cast. To clear quorum it needed roughly ten times the weight it had drawn in six days; in the last day it drew sixteen, and every GNO of the increase went to for – the against column finished on the same 115.27 it carried a day out. Both the thin tally and the concentration were properties of the reading date rather than of the proposal, which is the hazard set out below and the reason this page now records closes rather than standings.

What GnosisDAO's turnout record actually shows

GIP-153's thin mid-window ballot invited an obvious reading – that GnosisDAO cannot muster its own quorum – and neither the record nor GIP-153's own outcome supports it. Every closed proposal in the gnosis.eth space was read off Snapshot's public API on 18 August 2026 and scored against the 75,000 GNO bar each one carried. Of the 62 genuine governance proposals closed under that quorum, 44 cleared it – 71% – at a median turnout of 80,807 GNO. The rate has improved rather than decayed: 20 of 31 in 2024, 18 of 23 in 2025, 6 of 8 so far in 2026.

The denominator is where this measurement usually goes wrong. 149 closed proposals in that space carry the 75,000 GNO quorum, but 87 of them are not governance. They are airdrop-phishing posts spammed into the space under a dozen or so templates – “Mystery Box AIRDROP | Gnosis x Manta”, “Official AIRDROP | Gnosis x Starknet”, “Official NFT (Gift BOX)” – of which 38 drew no votes at all and the median drew one. Counting proposals straight off the API returns “44 of 149, or 30%, clear quorum” at a median turnout of 1 GNO, and would have confirmed the pessimistic reading. The numerator is identical under both counts, because no spam entry has ever come near quorum; the whole difference is a denominator inflated 2.4x by items that were never votes. A permissionless proposal surface is a spam surface, so any participation statistic computed over one without that filter is measuring the spam – a hazard worth carrying into DAO metrics and analytics and voter apathy, where turnout rates are the headline number.

The GIP-153 reading above was taken mid-window, and the window is where it changed. GIP-152 closed on 12 August with 108,507 GNO from 49 voters, 145% of quorum, and GIP-153 opened into the same electorate the same afternoon. Six days in, on 18 August, it had drawn 7,775 GNO from 34 – about 10% of the bar – which is exactly the thin ballot the pessimistic reading predicts. It closed the following afternoon, 19 August 2026 at 12:55 UTC, with 123,424.71 GNO from 54 voters, 165% of quorum (final scores read from Snapshot's GraphQL API on 3 September 2026). Roughly 94% of the eventual turnout arrived in the last day of a seven-day window.

That is a measurement hazard, not a GnosisDAO quirk. A quorum reading taken while a vote is open is not a smaller version of the result: on a fixed-length Snapshot window, large delegated weight tends to be cast late, so a mid-window tally measures how far through the window the reading is at least as much as it measures interest. The failure mode a directional vote expiring below quorum does produce – no recorded answer either way – is real and set out at quorum and threshold design; GIP-153 simply is not an instance of it. Any turnout statistic assembled from open proposals inherits this bias, which is a second hazard for DAO metrics and analytics alongside the denominator problem above.

A fourth proposal ran alongside it and was a different kind of business entirely: GIP-156, opened 17 August and closed 24 August 2026, asked whether a named participant should be barred from GnosisDAO's governance and communication channels. It finished at 23,368.84 GNO on 20 votes – 31% of quorum, with 99.8% of that weight voting For. So the question drew a near-unanimous answer and still expired with no recorded verdict, which is the clearest illustration on the docket that the 75,000 GNO bar was calibrated for treasury decisions rather than for moderation: a moderation question can be uncontested and still fall short of a threshold sized for spending.

What has closed since the census. Four genuine governance proposals closed in the fortnight after the 18 August reading, all four read final on 3 September 2026: GIP-153 cleared quorum, GIP-156 did not, GIP-154 – whether GnosisDAO should fund Gnosis Ltd at 15m a year – closed 27 August with 104,173.34 GNO from 44 voters at 139% of quorum, and a signal proposal to revive GIP-99 closed the same day with 5,145.21 GNO from 20 voters, 6.9% of the bar. Added to the census on its own filtered basis, 2026 becomes 8 of 12 clearing quorum and the whole-history figure 46 of 66, or 70%.

How Caper approaches this

GnosisDAO had to hold a contentious governance fight to build a treasury redemption at all — and even then it is a one-time, opt-in event. Caper makes that redemption a standing, protocol-native right. Every caper's exit() function lets a member hand back their vote tokens together with the governance tokens they hold and redeem a slice of the treasury at any time — sized, unlike GIP-151's, by weight rather than by balance — no proposal, no quorum, no bespoke smart contract. Crucially, a member's redemption share equals their canonical vote weight (t·v)/(V·T), computed by the exact same function that sets their voting power: the influence you carry and the value you can walk away with are one and the same number. Where GnosisDAO's saga asked "should the token be a claim on the treasury?", Caper answers that at the protocol layer — the "cash-out button" is always present, and it is inseparable from governance rather than voted into existence.