HIP-3 builder-deployed perps are permissionless perpetual markets that inherit HyperCore's onchain order book and margining. A qualifying deployer stakes 500,000 HYPE, the official mainnet requirement as of August 2026, to list one independent perp dex.
That 500,000 HYPE figure is the protocol's own staking requirement, and the same spec says it is expected to decrease as the infrastructure matures. The coins are a deployer lock against bad market operation. They are not a pool that pays traders if a listing goes badly.
Permissionless listing and a slashable stake sit on the same product.
What HIP-3 builder-deployed perps are
HIP-3 perps inherit HyperCore, so orders, cancels, liquidations, and margin live on the same onchain book that already runs validator-operated perps. A builder listing is a new dex on that book. The matching engine stays onchain.
A generic primer on perpetuals versus futures still helps if funding and mark price are new. It will not tell you where HIP-3 actually clears. HyperCore is the venue: "HyperCore includes margin and matching engine state." (source: Hyperliquid Docs, HyperCore overview)
What Hyperliquid is splits HyperCore from HyperEVM. HIP-3 lives on the book half. A contract call that burns HYPE is a different environment from a HIP-3 fill, even though both wait on the same HyperBFT order. How that book sequences cancels against fills sits in how HyperCore works.
| Validator perp | HIP-3 dex | Offchain | |
|---|---|---|---|
| Lists | Validators | Staked deployer | Hosted operator |
| Matches | Onchain book | Same book | Offchain |
| Margin | HyperCore | Per dex | Operator |
| 500k stake | No | Deployer stake | No |
Fills use the same trading API as the rest of HyperCore. The public Python client the project points at is an SDK for Hyperliquid API trading with Python (source: hyperliquid-python-sdk). That is a trading client, not a listing wizard. A page that walks perpDeploy JSON as the way to understand the product has switched audiences.
A HIP-3 market can look exotic, a metal, an equity ticker, a custom index, and still clear like the rest of the book. The exotic part is who listed it and who posted the oracle. The familiar part is HyperCore.
The 500,000 HYPE listing stake
The stake is how HIP-3 is secured at the listing layer. Mainnet still names 500k HYPE as the current official requirement as of August 2026. The spec's next clause is that this requirement is expected to decrease as the infrastructure matures. Treat that pair as a dated official line, not as a protocol constant you can bake into a model.
The spec's next clauses sit around that line. The staking requirement is maintained for a minimum of 183 days after the dex is deployed. Any deployer that meets the requirement can deploy one perp dex. Each perp dex has independent margining, order books, and deployer settings. "The slashed stake by the deployer is burned instead of being distributed to affected users." Even if the deployer has unstaked and started a staking withdrawal, the stake stays slashable during the 7-day unstaking queue (source: HIP-3: Builder-deployed perpetuals). The 183-day period starts at dex deploy, not at your first fill. The one-dex rule is current policy, with a documented future upgrade that might share a stake across multiple dexes. Independent margining means a blow-up on one builder book is not automatically the same margin account as validator-operated BTC.
Suppose you hold HYPE on an exchange and then see a builder-deployed silver perp on HyperCore. The 500k is the deployer's listing stake. Validators can slash it by stake-weighted vote if market operation goes badly. Slashed HYPE is burned. Your position still lives or dies on mark, margin, and the oracle that deployer set. The stake raises the cost of listing garbage. It does not write you a check.
How that stake works as a HYPE-holder mechanism, including what you forgo while 500k sits there, is the job of stake-to-deploy on Hyperliquid. The figure, the hedge, and the 183-day floor are the listing facts. Delegation clicks, and any promise that 500k never moves, are not.
500k HYPE is official today. The spec already says it should fall. A page that treats 500k as immutable is not reading the same sentence.
One deployer, one independent dex
A HIP-3 listing is its own perp dex, with its own order books, its own margin, and its own deployer settings. The person who posted the 500k can run one of those dexes under the current rule.
That independence is the holder-facing mechanic. You can be fine on validator-operated ETH and still get liquidated on a builder gold book if that dex's margin, oracle, or halt is the one that moved. A halt on the builder dex settles that dex's positions to mark. Validator-operated books keep trading. The first three assets on a new dex skip the shared Dutch auction. Extra listings join an auction that is shared across HIP-3 dexes. Those are listing facts, not a reason to open a deployer handbook.
Oracle choice is the deployer's job, and it is the job that slashing later inspects. The spec wants a well-defined underlying that is hard to manipulate. Most price indices, it says, are not amenable as perp oracles. You do not set that feed. You do have to notice who did, because every listed market on that dex sits under the same slashable operator.
Quote-asset choice matters for fees and for whether the collateral later loses quote status. HIP-3 deployers are not slashed for quote-asset failure in the current spec. A later upgrade might migrate a disabled quote, though the docs call that not expected on mainnet. For a trader, the practical read is narrower: know what you are margined in, because a builder dex can use a quote the main book does not. Getting that quote onto HyperCore is a Hyperliquid deposits and bridges job, not a HIP-3 listing right.
You do not need the auction hyperparameters. The shape is enough: one stake, one dex, separate books. If a UI pastes a builder market next to BTC without showing the dex name, ask which book you are on before you size it.
HIP-3 vs HIP-4 vs builder codes
Three builder labels sit on this chain, and they name different products. HIP-3 is levered perps on HyperCore with a slashable deployer stake. HIP-4 is a different primitive with bounded outcomes. Builder codes are a fee split on order flow, a separate product from a listing stake.
HIP-4 outcomes are fully collateralized and bounded. Outcomes bring non-linearity, dated contracts, and an alternative form of derivative trading without leverage or liquidations (source: HIP-4: Outcome markets). That is a different product from a HIP-3 perp with a new ticker. The product walk lives in HIP-4 outcome markets. The compare for who can lose more than collateral sits in how HIP-3 differs from HIP-4. Permissionless HIP-4 deployer actions are still labeled testnet-only in the docs index (source: Hyperliquid Docs, llms.txt). A live outcome market is not proof that anyone can permissionlessly deploy one on mainnet today.
Builder codes are a different lever again. Builder codes let builders receive a fee on fills they send on behalf of a user (source: Hyperliquid Docs, builder codes). The user approves a max fee. Codes are not a 500k listing stake and not a new dex. The fee-on-flow product sits in builder codes on Hyperliquid.
| HIP-3 | HIP-4 | Builder codes | |
|---|---|---|---|
| What | Builder perps | Bounded outcomes | Fee on fills |
| Leverage | Yes | No | Not a market |
| 500k stake | Yes, can fall | No | No |
| Mainnet | Staked deployer | Testnet-only API | Per-order |
If you came for builder perps, HIP-3 is the product. If you came for a yes/no market you wanted HIP-4. If you came for app fee share you wanted builder codes. Mixing those three is how a 500k figure lands on the wrong product.
HIP-3 fees and growth mode
HIP-3 trades sit in the same user fee tier as other perps and spot. Deployers can take an extra cut on top of that tier. The official range is a configurable share, not a frozen half-the-fee slogan.
HIP-3 deployers can configure an additional fee share between 0-300% (0-100% for growth mode). When a deployer activates growth mode, there is a 90% or greater reduction on the all-in fees. The markets must be entirely disjoint from existing validator-operated perps to prevent parasitic volume (source: Hyperliquid Docs, fees). If the share is above 100%, the protocol fee rises to match. Growth mode also cuts rebates and volume contribution by at least 90% while sitting on top of staking discounts. A 50% keep line exists later on the same fees page for spot and HIP-3 deployers. That is a different clause from the 0-300% share. Do not collapse them or bake a 97% split that is not on the official fees page.
Growth mode is the holder-facing fence. Cheap fees on a builder book are a mode with eligibility rules, a 30-day cooldown per asset, and a validator vote that can disable an ineligible listing. Crypto perps against any collateral are the obvious fail. The full ineligible list is a FAQ, because the rule you need in the body is the disjoint test, not an operator matrix.
Some of the remaining fee stream still hits HLP and vaults, plus the assistance fund and deployers. A HIP-3 fill is not a company-treasury event on the official fees page. Live APY on a vault is a different question and a moving figure.
Check whether the market in front of you is growth-mode cheap because it is disjoint, or regular-fee because it overlaps the main book, then check the deployer share. Those two knobs move what you pay. They do not move the 500k stake.
How HIP-3 deployer slashing works
Slashing on HIP-3 is a deployer penalty. Validators run a stake-weighted vote on the deployer. The coins that come out are burned. They are not distributed to people who lost money on that builder market.
That is the tension a holder has to sit with. The same stake that makes listing expensive does not make a trader whole. Malicious operation, incompetent operation, and a compromised deployer key are not distinguished. The spec cares about the effect on protocol correctness, uptime, and performance. Attempted bad inputs can still be slashable. Ordinary protocol bugs that are not tied to irregular deployer inputs are out of scope. Subjective dislike of a market is not a protocol slash, though downstream apps and reputation can still react.
Cross margin across dexes with different deployers is extra risk even when the system math still solvency-checks. Enabling cross margin on an asset is irreversible. Mainnet validators are supposed to keep that switch on books that have observable liquidity, a reliable external oracle, and resistance to manipulation. A 50% daily move versus the start-of-day price triggers a review. Assets that are expected to do that more than once a month are not supposed to get the switch. None of that is a click path. It is the reason a cross-margin HIP-3 book is a different product from isolated-only junk.
The trader-facing residual, including why LST operators have to diligence deployers, is covered in HIP-3 deployer slashing risk. Deployer stake can be slashed and burned. Your position is still your position. A halt settles to mark. It does not issue an insurance payout.
The deployer, and anyone whose coins sit in that stake, can lose the slashable HYPE, including through the 7-day unstaking queue. The trader on that dex does not receive the burned coins. An LST depositor behind a deployer takes the same slash if the operator staked those coins. A BloFin HYPEUSDT holder still has ordinary perpetual PnL on that listing, and none of the HIP-3 deployer stake.
You do not need the 100% / 50% / 20% guideline ladder memorized. You need the direction: slash is a burn, and it is not a trader hedge.
HIP-3 listings and BloFin HYPEUSDT
HIP-3 is permissionless builder perps on HyperCore's onchain book. The current official mainnet stake is 500k HYPE, it is expected to decrease, and it has to be maintained at least 183 days after the dex is deployed.
Each listing is its own dex. HIP-4 and builder codes are other products. Growth-mode fees are a disjoint-market privilege, not a HIP-3 default. Slashing burns deployer stake instead of paying you.
The live HYPE surface on BloFin is HYPEUSDT, a USDT-margined SWAP at 75x, listed December 19, 2024 11:30 UTC. Spot HYPE/USDT listed May 30, 2025 13:30 UTC. In the public instruments JSON the swap key is HYPE-USDT (source: BloFin, SWAP instruments). Those identifiers are venue instrument keys, not HIP-3 market IDs. HYPERUSDT on BloFin is Hyperlane, a second collision that has nothing to do with builder perps. A BloFin HYPE position does not put 500k into a deployer stake, and it does not give you a position on a builder silver book.
CoinGecko still describes Hyperliquid as a decentralized perpetual futures order book exchange on its own L1 (source: CoinGecko, Hyperliquid futures). That native book is a competing venue. Naming it is not a ranking. If you want HYPE exposure at BloFin you are on the CEX listing. If you want a specific HIP-3 market you are on HyperCore with that dex's oracle and that dex's stake.
You do not need to memorize fee-share percentages, auction reserve math, or every slash guideline. Weigh the book you are actually on, who posted the oracle, and whether the 500k in the spec is being treated as a living official number or as a slogan. Then decide whether that is a market you wanted, or whether you only wanted HYPE.
Frequently asked questions
If the 500k requirement falls, can a deployer unstake the extra HYPE?
Yes. The spec says any amount staked above the most recent requirement can be unstaked, so the 183-day maintain rule is a floor after dex deploy, not a ban on trimming excess once the official number moves down. Extra above the new requirement is withdrawable under that clause. The remaining required stake stays slashable, including through a 7-day unstaking queue. A drop in the published requirement is not a promise that every deployer will unstake, and it is not a change to your open position.
Which HIP-3 markets are ineligible for growth mode?
Crypto perps against any collateral are the clean fail, along with crypto indexes, ETFs, baskets, and linear mixes that still contain crypto. Gold is called out because PAXG-USDC already tracks gold. The list is illustrative, and validator vote is the last word the same way delistings work. Growth mode also needs a deployer fee scale between 0 and 10 and carries a 30-day cooldown per asset, so a cheap HIP-3 taker rate is gated to books that stay off the validator set.
Are HIP-3 deployers slashed if the quote asset fails?
No. Quote-asset failure is a different bucket. HIP-3 deployers are not slashed for a quote token that later loses quote status. The spec treats that as a fee and product issue, because quote-token issuers carry their own stake and slash conditions. Docs treat moving a disabled quote onto a new collateral token as a possible future upgrade, and they say that path is not expected on mainnet, so your residual is still the collateral you posted on that dex.
What happens to my position if the deployer halts the market?
Open orders cancel, and positions settle to the current mark. The same action can resume trading, which is how a deployer recycles an asset to list a dated contract without paying the shared auction again. Once every asset on that dex is settled, the required stake is free to unstake, subject to the 183-day floor and the 7-day queue. Halt is settlement. Validator-operated books on HyperCore keep trading.
Are builder codes the same product as HIP-3?
No. Builder codes pay an app a fee on fills it sends for you after you approve a max fee from your main wallet, with caps of 0.1% on perps and 1% on spot and at most ten active approvals. The builder also needs at least 100 USDC in perps account value. None of that is a 500k listing stake, and none of it creates a new perp dex, so you can see builder-code fees on a validator-operated market and never touch HIP-3.
Do HIP-3 fills sit in a separate fee tier from validator-operated perps?
No. Each user has one fee tier across perps, HIP-3 perps, and spot, with spot volume counting double toward that tier, sub-account volume counting toward the master, and vault volume not counting. Maker rebates pay the trading wallet on each fill. Growth mode is the exception that cuts protocol fees, rebates, volume contribution, and L1 rate-limit contribution by 90% on that HIP-3 asset.
Does a HIP-3 dex have its own backstop liquidator?
Yes. Each HIP-3 dex has a fully onchain backstop strategy, and it only takes over backstop-liquidatable positions from that dex. It only accepts assets where cross margin is enabled, which is why those assets are less likely to hit auto-deleveraging in a spike. That liquidator is an independent user on the book that still falls back to ADL to keep the dex solvent. There is no user action, so you do not opt in.
Researched and written by the BloFin Academy editorial team with AI-assisted drafting. Updated August 2026. Primary sources include the Hyperliquid HIP-3 and HIP-4 specs, HyperCore overview, fees, builder codes, and docs index, plus CoinGecko's Hyperliquid futures page, BloFin's public SWAP instruments API, and the public Python trading SDK. Protocol facts independently verified against cited sources current as of August 2026.
This article is educational and general in nature, not financial or investment advice. Cryptocurrencies like HYPE carry real risks, including price volatility, deployer and oracle risk on builder markets, slashing of deployer stake, venue risk, and the chance of losing funds. Nothing here is a recommendation to buy, sell, hold, deploy, or trade any HIP-3 market. Do your own research, and consider speaking with a licensed professional before making financial decisions. BloFin does not provide investment advice.