Both non-custodial perpetual DEXes. Different architectures: Hyperliquid runs a pure on-chain orderbook on its own L1; GMX V2 uses hybrid orderbook + multi-asset pool settlement on Arbitrum. Different funding, different LP economics, different sweet spots.
Both Hyperliquid and GMX V2 are non-custodial perpetual-futures venues. The decision usually comes down to architecture: a pure orderbook (HL) vs a hybrid orderbook + pooled-liquidity model (GMX V2). Each has different implications for spreads, LP economics, and which strategies fit best.
Numbers reflect both protocols’ states as of May 2026. Both adjust fees, asset lists, and pool parameters regularly; confirm at the venue before deploying.
| Dimension | Hyperliquid | GMX V2 (Arbitrum) |
|---|---|---|
| Custody | Non-custodial (HL L1) | Non-custodial (Arbitrum L2 smart contracts) |
| Order matching model | Pure on-chain CLOB | Hybrid orderbook + GM pool settlement[1] |
| Settlement chain | Hyperliquid L1 (HyperBFT) | Arbitrum One |
| Settlement asset | USDC (main) / USDH (Felix) | USDC + multi-asset GM pool tokens |
| Funding model | Hourly funding (longs ↔ shorts) | Funding fee (longs ↔ shorts) + borrow fee (paid to LPs)[2] |
| Max leverage (BTC perp) | Up to 50x | Up to 100x |
| Trade fees | 0.015% maker / 0.045% taker (Tier 0) | 0.05–0.07% open/close (no maker/taker split)[2] |
| Asset list | ~150 crypto perps + HIP-3 equity perps | ~30 crypto perps (curated) |
| LP token / vault | HLP vault (earns fee share + funding) | GM tokens (multi-asset pool position) |
| KYC at venue | None | None |
Both products adjust terms over time. Confirm live terms at each venue before trading.
Keel is a Strategy OS for AI-assisted systematic trading on Hyperliquid. Backtest, optimize, and run live strategies — realistic fees, slippage, and funding modeled.
Hyperliquid runs a pure on-chain central limit orderbook (CLOB) — same model as a traditional exchange, with matched buyer/seller pairs. GMX V2 is a hybrid: orderbook for triggers, but settlement against the GM pool (a multi-asset liquidity pool, similar to GLP). This affects price impact, slippage, and how funding/borrow fees work. Pure CLOB tends to have tighter spreads on liquid pairs; pooled liquidity handles smaller assets better.
HL uses a standard perp funding mechanism — hourly settlement, rate computed from perp-vs-oracle divergence. GMX V2 splits the cost into a funding fee (between longs and shorts, like HL) plus a borrow fee (paid to LPs for using the pool's liquidity). The borrow fee is GMX-specific and exists because perps consume the pool's collateral capacity. Net cost to traders is comparable; the structure differs.
HL is usually the answer for HFT — pure CLOB matches the model HFT firms have built infrastructure around, and HL's on-chain finality is fast enough for most strategies. GMX V2 is better suited for moderate-frequency directional traders who want pooled liquidity and don't need microsecond execution.
On HL, the HLP vault provides liquidity to the orderbook and earns from spread + fee share + funding pass-through. On GMX V2, GM tokenholders provide liquidity to the multi-asset pool and earn borrow fees, funding-rate spread, and LP fees from swaps + trades. Both have similar economics: LPs are net-counterparty to trader PnL on top of fee income. Profitable trader strategies = losses for LPs on the position side; unprofitable trader strategies = bigger LP gains.
Both are non-custodial — your funds stay in wallets you control. HL runs on its own L1 with HyperBFT consensus; GMX runs as smart contracts on Arbitrum One. HL has tighter integration between orderbook and settlement; GMX is more composable with Arbitrum DeFi. "Decentralized" depends on what you weight — single-purpose L1 vs general-purpose L2.