Compare Layer 1 and Layer 2 blockchains on fees, speed and security with September 2026 metrics from Ethereum, Solana, Arbitrum, Base and others.
Independent Layer 1 blockchains such as Ethereum, Solana, BNB Chain, and Tron operate as self-contained base layers. Each handles its own consensus, security through dedicated validator sets, and final settlement without relying on another chain. Understanding layer 1 vs layer 2 differences is essential when evaluating fees, speed, and security.
Layer 2 solutions, mainly optimistic and zk-rollups on Ethereum such as Arbitrum, Base, and Optimism, take a different approach. They execute transactions off-chain in batches, then post compressed transaction data or validity proofs to the L1 for settlement. This design lets L2s inherit security from the L1’s economic incentives and validator set while reducing on-chain load.
L1s therefore act as the primary settlement layers and continue to hold the majority of tracked liquidity and stablecoin supply. L2s achieve higher throughput and lower fees through batching, though they introduce trade-offs such as sequencer centralization risks or dispute windows on optimistic rollups. As of 14 September 2026, L1s accounted for 96.7 percent of tracked TVL across the ten networks measured by DeFiStar.io.
As of 14 September 2026, live estimates from the YFarmX gas fee checker show stark differences in cost between Ethereum L1 and its L2s, plus other major chains. Ethereum L1 ERC-20 transfers carried an estimated fee of $0.037821. In contrast, the optimistic rollups reported execution-only costs well below that threshold: Arbitrum One at $0.003327, Base at $0.000998, and Optimism at $0.000166. Solana modeled transactions at 200k compute units cost $0.000519, while BNB Chain ERC-20 transfers sat at $0.002357.
Execution-only figures for Arbitrum, Base and Optimism exclude the cost of posting compressed data or proofs back to Ethereum L1. Full costs on those networks therefore rise once L1 settlement is included, although the YFarmX snapshot isolates the off-chain execution component for direct comparison. Solana and BNB Chain figures reflect complete on-chain settlement within their respective base layers.
| Network | Type | Fee Estimate (USD) | Notes |
|---|---|---|---|
| Ethereum | L1 | 0.037821 | Full ERC-20 transfer cost |
| Arbitrum One | L2 | 0.003327 | Execution-only |
| Base | L2 | 0.000998 | Execution-only |
| Optimism | L2 | 0.000166 | Execution-only |
| Solana | L1 | 0.000519 | Modeled @200k CU |
| BNB Chain | L1 | 0.002357 | Full ERC-20 transfer cost |
These snapshots illustrate why users route high-frequency activity to L2s or alternative L1s, yet the gap narrows once L2 withdrawal or cross-chain bridging fees are factored in. Median non-vote fees on Solana reached roughly $0.000555 in a separate 16:33 UTC observation the same day, confirming the sub-cent range across non-Ethereum L1s.
As of 14 September 2026, DeFiStar.io data shows Base achieving roughly 185 TPS with 2-second blocks, Arbitrum at 29 TPS with 0.25-second blocks, and Optimism at 21.5 TPS with 2-second blocks. These DeFi-context figures place several L2s ahead of most L1s, where BNB Chain records about 100 TPS at 3 seconds, Avalanche 50 TPS at 2 seconds, and Ethereum 15 TPS at 12 seconds.
Solana’s report on the same date lists 4,685.27 total TPS, with non-vote transactions near 2,517 and a mean slot time of 0.32 seconds. The gap between total and non-vote TPS illustrates how consensus votes inflate raw counts on high-throughput L1s.
Upgrades continue to shift these numbers. Ethereum’s Fusaka upgrade from December 2025 raised blob capacity and gas limits, sustaining L2 throughput gains. Solana activated 350 ms slot times toward a 300 ms target, with the pending Alpenglow upgrade targeting roughly 150 ms finality.
Across the tracked set, L2s average 50.7 TPS versus 42.5 TPS for L1s, though Base has captured a growing share of daily activity volume.
Layer 2 networks derive their core security from the validator set and economic incentives of the underlying Layer 1. Transactions executed on an L2 are ultimately settled on the L1, so any attempt to rewrite history must overcome the L1’s consensus rules and staked capital. This inheritance allows L2s such as Arbitrum and Optimism to post compressed data or validity proofs to Ethereum without maintaining their own validator pools.
Optimistic rollups introduce additional risks during the dispute window, typically seven days, when any party can challenge a fraudulent state transition. Users who want to withdraw funds must therefore wait out this period or rely on liquidity providers. ZK-rollups replace the window with cryptographic proofs but still depend on the L1 for data availability and final settlement.
Sequencer centralization remains a practical concern. Many L2s currently route transaction ordering through a single operator that can censor or reorder activity until a permissionless replacement is deployed. In contrast, native L1s such as Ethereum and Solana secure every block directly through their own validator sets and token-weighted incentives, without an intermediate operator layer. The trade-off for L2 users is therefore speed and cost against a reliance on both L1 integrity and the operational honesty of the sequencer during the interim period before full decentralization.
As of 14 September 2026, L1 settlement layers hold the overwhelming share of on-chain liquidity. DeFiStar.io data records L1s controlling 96.7 percent of the $14,066.9M tracked TVL across ten networks, with Ethereum alone at $13,226.8M. Even modest L1s such as Avalanche ($63.4M) exceed several L2s; Arbitrum reports $204.7M and Base $200.2M.
Stablecoin supply shows the same concentration. Stablecoin Beat figures list Ethereum L1 at $148.6B (48.45 percent of the $306.7B total), Tron L1 at $94.2B (30.69 percent), and BNB Chain L1 at $17.5B. Base L2 and Arbitrum L2 together account for only $9.1B (2.96 percent combined).
Because L1s furnish final consensus, security, and settlement that every L2 must post to, the majority of both TVL and stablecoin issuance remains anchored on the base layers.
L1s like Ethereum reach finality once blocks are confirmed by validators, typically within one or two block times. L2s inherit security from the L1 but add delays: optimistic rollups impose a 7-day dispute window before withdrawals finalize, while zk-rollups settle faster once proofs are verified on the L1.
Bridges introduce smart-contract and sequencer risks. Funds can be frozen or lost if the bridge contract is exploited or if the L2 sequencer acts maliciously before L1 settlement occurs. Users should verify bridge audits and withdrawal mechanisms before transferring large amounts.
Ethereum’s Fusaka upgrade in December 2025 increased blob capacity and gas limits, contributing to sustained low fees on both L1 and L2s as of September 2026. Observed ERC-20 transfer costs dropped to roughly $0.000166–$0.003 on leading L2s, though spikes can still occur during congestion.
Choose an L2 for frequent DeFi trades or transfers when sub-cent fees and higher TPS matter most. Base recorded ~185 TPS and Optimism ~21.5 TPS on 14 September 2026, versus Ethereum L1 at ~15 TPS. Use an L1 when you need immediate finality or maximum liquidity depth without bridge exposure.
