Aptos Tests MonoMove With Up to 55x Faster Execution

Aptos Labs is developing MonoMove, a rebuilt execution engine and Move Virtual Machine that produced substantially faster smart-contract execution in internal testing. Selected Decibel transactions executed as much as 55 times faster than on Aptos’ current Move VM, although those results remain benchmarks rather than production mainnet performance.

In its official October 6 announcement, Aptos described MonoMove as the largest scalability upgrade in the network’s history. The company replayed real mainnet transactions from Decibel, an on-chain trading application, through both execution engines. Collateral withdrawals produced the largest measured improvement at up to 55x, followed by vault requests at 40x, perpetual-market requests at 38x and order placement at 22x.

Those figures measure execution of specific workloads rather than whole-network throughput. In separate benchmarks covering trading, lending and payment workloads against a real database, Aptos Labs reported an end-to-end throughput increase of up to 11x. The 11x result was achieved before enabling MonoMove’s planned parallel execution through Block-STM, according to the company.

Introducing MonoMove, Aptos' new execution engine and the biggest scalability upgrade in network history.

Moving global markets onchain takes a new standard of speed and safety. MonoMove delivers both, executing the smart contracts behind every onchain market up to 55x faster.… pic.twitter.com/qeAdL73Zfx

— Aptos (@Aptos) October 6, 2026

MonoMove Rebuilds the Execution Layer

MonoMove redesigns the machinery that executes Move smart contracts rather than changing consensus or simply reducing block intervals. The new implementation uses specialized execution techniques intended to remove overhead from repeatedly interpreting generic Move bytecode. Its purpose is to make contract computation itself faster, particularly for complex financial applications where execution can become a bottleneck.

That makes MonoMove distinct from Aptos’ other recent performance work. The network has separately pursued shorter confirmation intervals through its Archon architecture and 25-millisecond block-time roadmap. Faster block production affects network latency, while MonoMove targets the computation performed after transactions reach the execution engine. Improvements at one layer do not automatically translate into equivalent gains across the entire transaction lifecycle.

Safety is also part of the redesign. Aptos Labs says MonoMove preserves Move’s existing rules governing how assets can be created, copied and transferred while adding redundant checks inside the verifier and VM. The company is running differential tests that execute identical workloads on the current engine and MonoMove and compare their results. Aptos also plans extensive fuzzing and formal verification of critical components before production deployment.

Development is already visible in the public aptos-core repository, where MonoMove components cover the runtime, loader, transaction executor, native functions and testing infrastructure. That makes MonoMove an active implementation project rather than a performance concept awaiting initial development, but its presence in the codebase does not mean validators are currently using it on mainnet.

Mainnet Deployment Remains Planned for 2027

Aptos Labs is targeting feature completeness by the end of 2026 and a mainnet rollout during 2027. Production activation remains subject to an Aptos governance vote, and the project has not announced a specific activation date.

That distinction matters because Aptos has simultaneously shipped other infrastructure that is already operational. Its Encrypted Mempool went live on mainnet in October, providing optional pre-execution confidentiality for transaction payloads to reduce front-running and sandwich opportunities. MonoMove remains at a different deployment stage.

Decibel provides a particularly relevant benchmark because its fully on-chain order book generates repeated execution-heavy operations such as order placement, collateral changes and market requests. Replaying those real transactions makes the tests more representative than synthetic instruction benchmarks, but it still does not establish how MonoMove will perform across all applications or under production validator conditions.

For now, MonoMove demonstrates substantial execution headroom inside Aptos Labs’ testing environment. The meaningful milestone will come when the rebuilt VM survives its verification program, receives governance approval and produces comparable gains under live mainnet workloads.