Sablier (sablier.com) is a token-streaming protocol: instead of paying a lump sum, a sender locks ERC-20 tokens in a smart contract that releases them to the recipient continuously, second by second, over a set period. It is widely recognised as the first token-streaming protocol on Ethereum, and it is the reference primitive for the pattern DAOs reach for constantly — vesting a contributor grant, paying an ongoing salary, or distributing an airdrop that unlocks over time rather than all at once.

Money streaming: the idea

The concept predates the protocol. In November 2018 Sablier co-founder Paul Razvan Berg published ERC-1620, “Money Streaming” — a standard for representing continuous payments on-chain, where a locked balance is released to a recipient at a fixed rate per unit of time. Berg and co-founder Gabriel Apostu then built the first implementation, deploying it to Ethereum mainnet in December 2019. Sablier frames this as “real-time finance”: the passage of time itself is the settlement mechanism, so a stream needs no trusted intermediary to meter out the drip — the contract simply makes the vested portion withdrawable as the clock advances.

Lockup, Flow, and Merkle airdrops

The V2 architecture splits the protocol into distinct, single-purpose product lines:

• Lockup — finite-duration streams for vesting and vested airdrops. A stream can carry a cliff (nothing withdrawable until a date, then a jump) and be created cancelable or non-cancelable, letting a DAO choose whether it can claw an unvested balance back.
• Flow — open-ended, cancelable streams for recurring payments and grants with no fixed end, closer to a live payroll rail than a vesting schedule.
• Merkle Airdrops — on-chain claim campaigns where eligibility is proven against a Merkle root, so an instant or vesting airdrop can be distributed to thousands of wallets without the issuer pre-funding a stream for each one.

Because the contracts are non-upgradeable and permissionless, anyone can create a stream with any ERC-20 on any supported chain, and no admin can alter a stream after the fact.

Why DAOs use it

Streaming solves a governance problem, not just a payments one. A DAO that votes a contributor a twelve-month grant does not want to hand over the whole sum on day one — that is a trust and alignment risk. Streaming the grant means the recipient earns it as they keep working, and (with a cancelable stream) a governance vote can stop the unvested remainder if the relationship ends. The same logic covers contributor compensation, team and investor token unlocks, and airdrops designed to reward continued participation rather than an instant flip. By its own reporting Sablier has served over 297,000 users and 550,000+ streams, moving billions in cumulative volume — it sits alongside Safe and the rest of the tooling stack as core treasury infrastructure. The continuous, open-ended counterpart is Superfluid, which streams money per-second with no escrowed pot — the two are the escrowed-vesting and live-netflow poles of on-chain streaming.

Maintenance mode and the public-good ending

On 13 July 2026 Sablier Labs announced it was entering maintenance mode and halting active development. Berg pointed to a sharp Q1-2026 drop in usage as customers postponed token launches, and to AI-assisted coding making it cheap for competitors to clone the product, concluding there was no venture-scale business in onchain streaming. The move is instructive for anyone relying on DAO tooling: the company stopped, but the protocol did not. The immutable, permissionless contracts keep running with no operator, existing streams and vesting plans are unaffected, the hosted frontend is committed through June 2028 before becoming a community-run public good, and Sablier said it would accelerate the licence flip that turns its Business Source License contracts into GPL. Depending on a contract that outlives its company is exactly what “permissionless” is supposed to buy. The licence half of that promise, though, has not landed yet.

Where the licence actually stands

Sablier’s EVM contracts live in sablier-labs/evm-monorepo, which is where the older sablier-labs/lockup path now redirects, and the repository’s LICENSE.md still opens “The primary contracts are licensed under the Business Source License 1.1 (BUSL-1.1)”. BUSL is a delayed-open licence rather than an open-source one: it grants copying, modification and non-production use immediately, requires a commercial licence for production use, and converts to a named Change License on a Change Date. Sablier’s LICENSE-BUSL.md names GPL-3.0-or-later as the Change License and sets the Change Date to “the earlier of 2029-07-01 or the ISO 8601 date stored in the eth.sablier.license-change-date text record”.

That makes the promised acceleration an on-chain write rather than a commit, so no amount of reading the repository can confirm it. Read on 20 August 2026, the record is empty on sablier.eth (resolver 0x4976fb03c32e5b8cfe2b6ccb31c09ba78ebaba41, whose url and description text records return values normally, so the read path is sound) and empty on lockup.sablier.eth. With no date written, the fallback binds and the core contracts stay BUSL-1.1 until 1 July 2029. A DAO forking or self-hosting Sablier today is forking source-available code, not GPL code – a distinction that bites hardest in exactly the scenario the maintenance-mode announcement was meant to reassure people about, and one worth checking on any tool that claims to be winding down into a public good.

How Caper approaches this

Sablier answers “pay this out gradually over time.” Caper answers a different question — “should the treasury spend at all, and on what?” A caper's treasury releases funds only through proposals that clear a vote, survive the market window that a pass opens, and only then execute a discrete PAYOUT or INVEST action, so every disbursement is a governed, on-the-record decision rather than a continuous drip. The two are complementary: a caper could vote a grant and settle it as a single payout, or a payout could itself fund a Sablier stream for a contributor. Where they differ is emphasis — Caper puts the deliberation and the funding decision on-chain and auditable; Sablier puts the schedule of an already-agreed payment on-chain. Both share the same underlying bet: money movement is safer when the rails, not a person, enforce the terms.

References

• Sablier, “What is Sablier” (docs)
• Ethereum, ERC-1620: Money Streaming · original proposal thread
• Paul Razvan Berg, “Introducing Sablier V2”
• Sablier, Token Vesting · “Analyzing 500K+ Vesting Streams”
• Paul Razvan Berg, “Sablier Labs is Entering Maintenance Mode” (13 Jul 2026)