Search

Search benchmark pages, protocols, published runs and the docs

Rout
No measured results yetStatus as of 2026-11-03

Execution you can check on chain

How often a transaction lands in the slot that triggered it, how long the tail is, what each success costs, and every signature behind it. Fixed before the run. Recomputable with your own RPC.

Read the methodProtocols

Rout has not traded on mainnet. Until a run is published, every result below reads “not measured”. rout-next/src/domain/capabilities.ts:44-45 · rout-automation-ui@61469bf

What makes these numbers checkable

  • The clock starts on chain

    Every delay counts from an on-chain anchor slot, never from our own receive time.

  • Every attempt counts

    Refused, expired and failed intents stay in the denominator.

  • Signatures are published

    Each figure links down to the transactions behind it.

  • Fixed before the run

    Metrics and pass rules are hashed before the first intent.

  • You can rerun it

    The verifier recomputes everything from your own RPC.

Where we're focusing, in order

Each benchmark is useful even if the next one never runs, and needs less than the one after it.
  1. 1Shadow of a partner's own execution

    Something useful for a partner with no keys and no traffic, and the workload Rout would have to beat.

    Headline: Partner intents landed within window, Δslot tail, fail classes

    Not measuredPartner signatures, an rbench importer, a licence · None
  2. 2Protection gap

    Sizes the window any guard could close, from public data. If it's ~0 slots, no guard-speed claim is worth making.

    Headline: Slots and outside flow between threshold cross and exit

    Not measuredApproval · RPC credits at most
  3. 3Submission path

    The first number about Rout's own path. Its primary question is whether durable nonce helps or hurts landing.

    Headline: Nonce vs blockhash landing difference; double-landed must be 0

    Not measuredEngine E0–E4 · About 0.16 SOL plus rent (estimate)
  5. 5Guard response

    The protective exit, end to end on chain, structural delay included.

    Headline: Exits landed within window of the recomputed threshold cross

    Not measuredMainnet, E6, E8, budget · Sized after E0
  6. 6Exits under volatility

    Bounded exits against unlimited slippage.

    Headline: Exits landed within window; shortfall in bps

    Not measuredMainnet, E9, budget · Sized after E0
  7. 7Head-to-head guard

    The decision quantity for a guard integration.

    Headline: Landing difference vs the incumbent; cost ratio; 0 new fingerprints

    Not measuredShadow and guard response done, partner enrolment · Partner
  8. 8Block-0 inclusion

    Only if the engine can carry a bundle the partner built and signed.

    Headline: Create and creator buys in one slot

    Not measuredEngine answer E5 = yes · Partner

What you'd ask

The questions engineers at launch platforms and trading services ask first. Open one to see exactly how it's measured.

Published runs

All runs
No published run yet
The first runs that need no engine work are the shadow benchmark of a partner's own execution and the protection-gap study.
Run the verifier on your data

Every capability a benchmark would test, tagged by the strongest evidence we could inspect. 7 protocols are drafted; 0 finalized.