Is Interop Solved? It is an overview of the evolution of cross-chain transaction speed over the past 3 years using Jumper as a case study. The question of interoperability can be approached from many angles: the execution efficiency, their security (how often a bridge fails), and their execution speed. Here, I’m focusing on the third aspect: I analyzed the median transaction time for users over the past three years, based on the bridging solutions used on @JumperExchange (and Transferto before that). To do this, I looked at data from around ten bridges, across five typical route types: ➡️EVM L2 to EVM L2 (Arbitrum, Base, Optimism) ➡️Ethereum Mainnet to EVM L2 ➡️EVM L2 to Ethereum Mainnet ➡️Solana from and to EVM (Ethereum, Base, Optimism, Arbitrum) 1. L2 → L2 (Arbitrum, Base, Optimism) ➡️ Mint/Burn bridges @circle CCTP V1 has remained stable over time: ~18 min @glacislabs , which recently launched and aggregates several standards, brings this down to ~25–30s ➡️ Intent bridges @AcrossProtocol is the standout here: it went from 100s to 10s in 2023 and from 10s → <3s in just a year in 2024. @MayanFinance , @squidrouter , and @RelayProtocol launched more recently and show median times of 15–25s, but haven't improved further since their launch ➡️ Liquidity bridges @HopProtocol dropped from 60–70s → 30s @StargateFinance from 70s → 35–40s Same pattern for @symbiosis_fi No major improvement have been made since July 2024. 2. Mainnet → L2 This is where slowness starts to bite. ➡️ Mint/Burn bridges Same trend as L2→L2: Circle and Glacis are stable, but Glacis sees more volatility and slower times in this direction ➡️ Liquidity bridges Improvements across the board Stargate: 260s → <200s Symbiosis: <80s in the past 3 months Hop: Less recent data, but same downward trend ➡️ Intent bridges Slower here due to Ethereum finality constraints Across/Relay: ~5–25s Squid: ~40s Intent is still faster — but constrained by Ethereum finality 3. L2 → Mainnet This direction is faster than the opposite. ➡️ Mint/Burn Absolutely no change for Circle Glacis is 4–5x faster compared to Mainnet → L2 ➡️ Liquidity bridges Strong improvements from 2022 and 2024, interestingly performance still seems to improve in the last months. Hop: 100s → 30s Stargate: 100s → 40–50s Symbiosis: 100s → 45s ➡️ Intent bridges Across again leads: 100s → <7s Squid, Mayan: 20–40s Relay is the odd one out — recent uptick to 30s+ 4. EVM ↔️ Solana EVM → Solana Surprisingly, bridging to Solana is faster than bridging from Mainnet to an EVM L2 and similar to bridging from an L2 to Mainnet. Relay: ~5s Mayan: ~25s Slight increase over time, but still very fast overall Solana → EVM Mayan and Relay show quite similar performance. Allbridge is a surprise here: 10x faster in this direction vs. EVM → Solana My Takeaways From a speed standpoint, interop isn’t fully solved — and we've been plateauing for about a year. In my view, we’ll only get there once top-90th percentile cross-chain transactions consistently settle in under 2 seconds. It might be the case for intent-bridge soon but we're far from here for other bridge type. 1⃣ Across all routes, we observe a general decline in transaction times from 2022 to 2024. However, this trend becomes less apparent from mid-2024 onward. 2⃣ The decrease in transaction time is often driven by the launch of new bridges or new versions of existing bridges, rather than improvements to existing ones—except in the case of Across, Hop, or Symbiosis. In some cases, we even observe a slight deterioration in performance after the launch of a new bridge (e.g., on certain routes or with Stargate). Once a new bridge is live, we even observe time performance getting a bit poorer from time to time (relay on specific routes) or Stargate. 3⃣ Intent-based bridges are consistently the fastest option across all routes. 4⃣ By aggregating multiple mint/burn bridge standards, Glacis demonstrates that acceptable transaction times can still be achieved with this type of bridge. 5⃣ L2-to-L2 routes are by far the fastest, followed by L2-to-Mainnet, EVM-to-Solana, and finally Mainnet-to-L2. In the current landscape, it remains challenging to build a ‘chain abstraction’ tool that integrates Ethereum Mainnet, as users are likely to encounter noticeable latency.