Stargate, Across, and LayerZero get lumped together in "best bridge" lists, but they sit at three different layers of the cross-chain stack. LayerZero is a messaging primitive. Stargate is an application that runs on top of LayerZero. Across is an optimistic intent-based bridge with its own relayer network. Picking the right one depends on what you're moving, which chains you need, and how much you trust each security model.
​

This guide breaks down the architecture, fees, supported chains, security assumptions, and asset routing for each in 2026, so you can choose without guessing.
​

The fastest way to see the differences. Numbers reflect mainstream USDC routes between Ethereum and an L2 in May 2026; specialty routes vary.

LayerZero is not a bridge. It is a cross-chain messaging protocol. Apps deploy "Omnichain Applications" (OApps) that send arbitrary messages between chains, and LayerZero handles delivery via a set of Decentralized Verifier Networks (DVNs) and Executors that each app configures itself.
​

That configurability is the security story. Each OApp picks its own DVN set, threshold, and block confirmations. A conservative app might require 3 of 5 DVNs including Google Cloud and Polyhedra. A cheaper app might require just 1. LayerZero v2 launched DVN modularity in 2024, and by 2026 the network covers 70+ chains including Solana, Aptos, Sui, TON, and every EVM L2 of consequence.
​

You don't really "use LayerZero" as an end user. You use a bridge or app built on it (Stargate, Radiant, Pancake CAKE bridge, Tapioca). What matters is that the underlying transport is consistent, so if you trust the OApp's DVN choices, you trust the bridge.
​

Stargate is the flagship Omnichain Application on LayerZero. It runs unified liquidity pools per asset (USDC, USDT, ETH) that span every supported chain. Send USDC from Arbitrum, and the same pool pays out native USDC on Base. No wrapped tokens, no synthetic IOUs.
​

This is Stargate's "delta algorithm" guarantee: every transfer is fully backed by the destination pool at the moment it is initiated. If a pool runs dry on a hot route, the protocol routes through an intermediate hop or rejects the transfer rather than minting a wrapped placeholder.
​

Fees in 2026 sit at roughly 6 basis points (0.06%) on the largest routes, paid to liquidity providers. Stargate added Hydra in 2024 to support burn-and-mint for canonical assets like CCTP-USDC, which dropped fees on those specific routes closer to zero. Coverage spans Ethereum, every major EVM L2 (Arbitrum, Optimism, Base, Polygon, Avalanche, BNB, Linea, Scroll, Mantle), plus Aptos.
​

Across takes a fundamentally different approach. When you initiate a transfer, you deposit funds on the source chain and broadcast an intent. Relayers (anyone running a bot) front you the destination assets out of their own inventory, usually within seconds. They then claim repayment from the source deposit after the UMA optimistic oracle verifies the transfer was legitimate, a process that takes a few hours but happens entirely in the background.
​

The user experience: 2 to 30 second fills, often the fastest in the market on supported routes. The security model: UMA's optimistic oracle plus relayer bonds. A fraudulent fill can be challenged and slashed during the dispute window, so relayers have direct skin in the game.
​

Fees in 2026 average around 4 basis points on mainstream USDC routes, often lower than Stargate because relayer competition is fierce. The tradeoff is chain coverage: Across focuses on Ethereum plus 15 or so EVM L2s where relayer inventory is thick. Don't expect Aptos or Solana support.
​

Each model has a different trust assumption, and "most secure" depends on what you're optimizing against.
​

LayerZero's security is only as strong as the DVN set the app picks. A well-configured OApp with multiple independent DVNs and Google Cloud as a verifier is hard to attack. A lazy 1-of-1 configuration is not. Stargate uses a conservative DVN set by default, and its pools are audited and publicly tracked on DeFiLlama.
​

Across relies on UMA's optimistic oracle plus economic bonds. It has never had a bridge exploit since launch in 2022, and the design means a successful attack requires either compromising UMA itself or outrunning the dispute window with a fraudulent fill that no one challenges. Both are hard.
​

For pure stablecoin transfers where speed and fee matter most, both Stargate and Across have strong production track records. For exotic routes or non-EVM chains, Stargate (via LayerZero) is the only realistic option of the three. For developers, LayerZero gives you the most flexibility but the most responsibility for getting security configuration right.
​

All three protocols deliver native assets on the destination chain when configured correctly, but the mechanics differ.
​

Stargate's pools hold native USDC on every chain, so you always receive native (not bridged) USDC. The exception is older long-tail assets where Stargate may still use a wrapped representation. Check the asset page before sending large amounts.
​

Across pre-funds destinations with native assets via relayers, so deliveries are always native. The relayer eats the inventory cost and gets reimbursed later.
​

LayerZero itself does not handle assets. The OApp built on top decides whether to burn and mint (canonical), lock and mint (wrapped), or use a liquidity pool (native swap). When you bridge USDC via Stargate, you get native. When you bridge a long-tail token via a smaller OApp, you may get a wrapped version. Always verify the contract address on the destination.
​

Take a 10,000 USDC transfer from Arbitrum to Base in May 2026. Stargate quoted approximately 6 USDC in protocol fee plus 1 to 2 USDC in destination gas reimbursement. Across quoted approximately 4 USDC total, including gas. A direct LayerZero OFT-USDC transfer (no liquidity pool, burn-and-mint) cost roughly 1 to 2 USDC, but only works for OFT-enabled tokens.
​

For small transfers under 100 USDC, gas dominates everything. The named "fee" matters less than the destination chain's gas cost at that moment.
​

For the full route-by-route breakdown, see support/en/articles/15197981 and support/en/articles/15197982.
​

A simple decision framework based on what you're doing.
​

Pick Across if you're moving stablecoins between Ethereum and an OP-stack L2 (Optimism, Base, World Chain) or Arbitrum and want the fastest fill at the lowest fee. Across dominates L2 to L2 routes.
​

Pick Stargate if you need broad chain coverage including BNB Chain, Avalanche, Polygon, or Aptos, or if you're moving large stablecoin amounts where deep unified liquidity matters. Stargate is also the go-to for OFT-style native asset transfers where supported.
​

Build on LayerZero if you're a developer who needs a cross-chain primitive for an app and wants to pick your own DVN security set. End users rarely interact with LayerZero directly.
​

None of these is a wrong choice. They serve different needs and they will all still be around in 2027. For a broader framework on bridge selection, see PLACEHOLDER-how-to-pick-crypto-bridge.
​

Eco Routes aggregates across multiple bridges including Stargate and Across, plus CCTP, Hyperlane, and others, then picks the cheapest fast route per transfer. Instead of integrating each protocol separately, apps point at Eco's API and get the best available route automatically. See our roundup of the best crypto bridges in 2026 for context.
​

Fee and speed figures pulled May 2026 from each protocol's live quote interface: stargate.finance, across.to, and layerzero.network. Architecture details cross-referenced against each protocol's official documentation (stargateprotocol.gitbook.io, docs.across.to, docs.layerzero.network) and DeFiLlama bridge volume data. Chain coverage counts reflect mainnet deployments as of May 2026 and will change as new chains launch.
​