The cheapest crypto bridge in 2026 is not a single answer. It is a function of transfer size. A $200 move and a $200,000 move route through entirely different rails, and the protocol that wins at one notional often loses at the other. This article ranks the major routes by total cost (protocol fee plus gas plus slippage) and shows where each one breaks even. For a static head-to-head by chain pair, see /support/en/articles/15197981; the focus here is how cost behaves as size scales from $100 to $1M.
​

A crypto bridge's total cost is the sum of four components: protocol fee (a per-transfer charge or basis-point rate set by the bridge), source and destination gas, slippage (on liquidity-pool routes), and the time-discount (cost of capital while funds are inflight). Each scales differently with size.
​

Protocol fees split into two families. Messaging bridges like Circle CCTP, Hyperlane warp routes, and Stargate charge a near-fixed messaging cost that barely scales with notional. AMM-style pool bridges charge a basis-point rate that compounds with size. Intent-based protocols like Across charge a relayer-bid spread that compresses on deep routes.
​

Gas is the variable nobody quotes upfront. On Ethereum mainnet at 30 gwei, a burn-and-mint call runs $4 to $8. On Base or Arbitrum the same call is under $0.10. Slippage only appears on AMM-style routes. The time-discount matters most for slow native rollup withdrawals (7 days on optimistic L2s) where idle capital has an opportunity cost. Headline protocol fees are not the right comparison: gas dominates small transfers and the protocol fee dominates large ones.
​

The table below ranks the major route families by total cost at three reference sizes ($500, $5,000, and $50,000) for stablecoin transfers between deep liquidity chain pairs (Ethereum, Base, Arbitrum, Optimism, Polygon). Numbers are typical ranges from DeFiLlama bridge dashboards, Circle's CCTP fee schedule, and live quotes pulled in Q1 2026. Soften any single quote by 20-30% for chain-pair variance.
​

Three patterns. Hyperlane warp routes and LayerZero OFT are the cheapest fixed-cost rails because their fee is messaging price, not a percentage of value. CCTP wins on native USDC because Circle does not charge a protocol fee for the burn-and-mint primitive (Fast Transfer adds a route-dependent fee, typically 0 to 14 bps per Circle's fee schedule). Aggregator routes (Squid, LI.FI, Jumper) layer 5 to 25 bps on top of whichever underlying bridge they pick, so they trail at scale.
​

CCTP charges no fee for the burn-and-mint primitive. The user pays gas on the source chain to burn USDC, then gas on the destination chain to submit the attestation and trigger the mint. Fast Transfer in CCTP V2 adds a small finality fee (variable, typically 0 to 14 bps by route) on top of gas. For retail-size USDC moves the total is gas-dominated.
​

CCTP economics are chain-dependent. A Base-to-Arbitrum call costs cents because both chains have sub-dollar gas. An Ethereum-to-Base call can run $4 to $8 because the burn happens on mainnet. CCTP barely scales with notional: a $50,000 transfer costs essentially the same as a $5,000 transfer, which makes CCTP the asymptotically cheapest USDC route for large transfers. The trade-off is scope. CCTP only moves native USDC. Circle's CCTP documentation lists the 13+ supported chains as of Q1 2026. V2 Fast Transfer attestation cuts finality to roughly 8 to 20 seconds, down from V1's 13 to 19 minute window.
​

Intent-based bridges quote a single all-in price upfront and let a network of solvers compete to fill the order. Across is the canonical example. The user deposits on the origin chain, a relayer fills the order on the destination in seconds, and UMA's optimistic oracle settles the relayer's claim a few hours later. Across publishes a 0% protocol fee. The cost the user pays is the relayer's bid: destination gas plus a risk premium for the inflight period.
​

On deep stablecoin routes (USDC and USDT between Ethereum, Base, Arbitrum, Optimism), solver spreads have compressed to roughly 1 to 5 bps. That makes intent rails competitive with CCTP at moderate sizes and sometimes cheaper at sub-$1,000 because the relayer absorbs destination gas. On thinner routes spreads can run 20 to 50 bps. Solver competition is the lever: more solvers compress spreads, fewer solvers widen them. ERC-7683 is the cross-protocol intent standard aimed at sharing solver liquidity across platforms; Across's docs describe the architecture.
​

Bridge aggregators like LI.FI, Jumper, Squid, deBridge, and Socket route across many underlying bridges. They solve a real problem (you do not have to pick which rail to use) and they introduce a real cost (a markup on top of the underlying route).
​

LI.FI's default integrator fee is roughly 25 bps, configurable down to 5 to 15 bps for partner integrations. Jumper, which is LI.FI's consumer-facing frontend, uses the same routing engine with the same fee structure. Squid charges in a similar 10 to 25 bps range and specializes in Axelar GMP routes (especially Cosmos-EVM corridors). For a $50,000 stablecoin move, a 25 bp aggregator markup is $125, which is more than the underlying CCTP or Hyperlane fee would cost on its own.
​

The break-even logic is simple. If the aggregator picks a cheaper underlying route than you would have picked manually, the markup pays for itself. If you already know your route (CCTP for USDC, Hyperlane warp for an ERC-20, LayerZero OFT for USDT0), routing directly is cheaper. Aggregators win on long-tail routes and lose on the deep stablecoin majors.
​

Transfer size is the dominant variable in route selection. Small transfers (under $1,000) are gas-dominated, so the cheapest route is the one with the lowest source-and-destination gas, regardless of protocol fee. Mid-size transfers ($1,000 to $50,000) are where intent rails win on deep routes and CCTP/Hyperlane/LayerZero win on fixed-cost messaging. Large transfers ($50,000+) reward the fixed-fee rails because percentage-based fees compound.
​

The decision tree for USDC. Under $500 between two L2s: any intent rail (Across) or CCTP V2 Fast Transfer, both under $2. $500 to $5,000: CCTP V2 if both chains have low gas, intent rail if Ethereum mainnet is involved. $5,000 to $50,000: CCTP V2 wins on cost almost universally; intent rails win on speed. Over $50,000: CCTP V2 or Hyperlane warp routes; aggregators trail by 20 to 100 bps.
​

USDT has no native CCTP equivalent. USDT0 via LayerZero is the closest analog (fixed messaging fee, omnichain). For non-LayerZero USDT routes, Hyperlane warp routes or Stargate are typical. Tron USDT (the largest USDT supply by chain) bridges through Wormhole, Allbridge, or centralized exchange off-ramps; see support/en/articles/15291265.
​

AMM-style bridges price liquidity through a constant-product or similar curve, which means slippage rises with trade size. A $50,000 move on a thin pool can incur 20 to 100 bps of slippage on top of the headline fee. Messaging bridges (CCTP, Hyperlane, LayerZero) avoid pool-based pricing entirely, so they do not slip with size.
​

This is the structural reason CCTP, Hyperlane warp routes, and LayerZero OFT pull ahead at scale. They pass messages between chains and let canonical mint/burn or lock/unlock handle the asset movement. No pool means no slippage. Stargate is a hybrid: LayerZero messaging plus pooled liquidity for instant fills, so it carries a 1 to 6 bp LP fee plus minor slippage on large trades.
​

Routes that go through DEX aggregators in the middle (swap source to USDC, bridge, swap USDC to destination) compound DEX slippage onto bridge cost. For pure stablecoin moves this is irrelevant; for token-to-token cross-chain swaps it can dominate.
​

Three costs hide in bridge UX. First, destination gas top-up. Bridging USDC into a chain where you have no native gas leaves you stranded; some bridges (Squid, Jumper) auto-include a small gas drop, others force a second bridge for $1 to $5 of gas. Second, approval gas: ERC-20 approvals add a one-time cost ($0.10 to $5 depending on chain) that headline quotes often exclude.
​

The third is the time-discount on slow native bridges. Optimistic-rollup native bridges (Arbitrum, Optimism, Base) have a 7-day withdrawal challenge window. At a 5% annual cost of capital, that is roughly 10 bps for the week. For a $100,000 withdrawal, the implicit cost is around $100, more than a fast third-party route would charge. The native bridge looks free until you price the wait.
​

For operators routing real volume, the rational frame is total landed cost: protocol fee plus gas plus slippage plus the time-discount. Eco Routes is the intent-based router that aggregates these rails (CCTP, Hyperlane, LayerZero, and others) and picks the cheapest one per route per size. It is one option for operators who want a single API surface across the rails listed above, not a separate competing bridge.
​

Sources and methodology. Protocol fee models verified against Circle CCTP fee documentation, Hyperlane warp route docs, Across intent architecture docs, and Stargate public fee schedule. Total-cost ranges aggregated from DeFiLlama bridge dashboards in Q1 2026 and live route quotes; soften any single figure by 20-30% for chain-pair and gas-condition variance.
​