ChainBench · Flotapponnier · Sep 19, 2026 · Sep 19, 2026
@@ -89,12 +89,12 @@ source: https://github.com/ChainBench/OpenChainBench/tree/main/harnesses/bridge-
 # are documented in the methodology until they accrue enough samples to panel.
 metric_panels:
   - id: fee
-    label: Realized fee
-    metric: avg by (bridge) (ocb:bridge_fees_usd:avg_7d{region="eu-west"})
+    label: Realized fee (bps of ticket)
+    metric: avg by (bridge) (ocb:bridge_exec_realized_fee_bps:avg_7d{region="eu-west"})
     label_key: bridge
-    unit: usd
+    unit: bps
     higher_is_better: false
-    description: "Amount sent minus amount actually received on the destination chain, in USD, averaged over the last 7 days of executions. The fee a user really paid, read on-chain rather than from the quote."
+    description: "Ticket minus the value that actually landed on the destination chain, in basis points of the ticket so the $3 and $30 executions compare, averaged over the last 7 days of settled executions whose destination balance was read on-chain."
   - id: slippage
     label: Slippage vs quote
     metric: avg by (bridge) (ocb:bridge_quote_slippage_usd:mean_7d{region="eu-west"})

@@ -4,7 +4,7 @@ slug: bridge-realized-cost
 number: "263"
 title: "Bridge slippage vs quote: what actually settles versus what was quoted"
 seo_title: "Bridge execution slippage 2026: realized cost vs the quote"
-seo_description: "Live gap between a bridge quote and the real on-chain settlement for Mobula, Relay and LI.FI, plus the gas the user actually pays, measured on real executed USDC transfers across Solana, Base and Arbitrum from EU-West."
+seo_description: "Live gap between a bridge quote and the real on-chain settlement for Mobula, Relay, LI.FI and Near Intents, plus the gas the user actually pays, measured on real executed USDC transfers across Solana, Base and Arbitrum from EU-West."
 subtitle: The gap between what a bridge quotes and what actually settles on-chain, in USD, on real executed USDC transfers, plus our own gas, at $3 and $30, measured from EU-West.
 category: Bridges
 status: live
@@ -24,7 +24,7 @@ dimensions:
 seo_intro: |
   Every bridge quotes a cost. This benchmark measures whether that quote
   survives contact with the chain. We broadcast and settle real USDC
-  transfers through Mobula, Relay and LI.FI on a self-conserving triangle,
+  transfers through Mobula, Relay, LI.FI and Near Intents on a self-conserving triangle,
   then record the realized fee against the fee the quote promised. The
   headline is that gap, in dollars: a bridge whose settlement matches its
   quote sits near zero, a bridge that fills below quote reads positive, and
@@ -45,23 +45,25 @@ abstract: |
 
 methodology:
   - "Route: a self-conserving all-USDC triangle (Solana to Base, Base to Arbitrum, Arbitrum to Solana). Each bridge completes the full triangle before the next starts, so inventory returns to origin and only fees burn."
-  - "Slippage vs quote: realized fee minus quote-projected fee, in USD (bridge_quote_slippage_usd). Realized fee is amount sent minus amount actually received on-chain. Near zero means the settlement matched the quote; positive means it settled worse; negative means it beat the quote."
+  - "Slippage vs quote: realized fee minus quoted fee, in USD (bridge_quote_slippage_usd). Realized fee is the ticket minus what actually landed on the destination, read from the balance delta; quoted fee is the ticket minus the output the quote promised, one definition for every bridge. Near zero means the settlement matched the quote; positive means it settled worse; negative means it beat the quote."
+  - "Why not the providers' own fee fields: some mix in gas paid in the native token, which the realized fee (ticket minus landed) can never contain, so one bridge read a constant negative offset at every ticket size. Both sides of the subtraction are now defined against the ticket."
   - "All-in gas: our own on-chain gas to execute (approve plus deposit), measured as the source-chain native balance delta before and after (bridge_exec_gas_usd). This is the cost the user bears on top of the bridge fee."
-  - "Realized output: the USD value that actually landed on the destination chain (bridge_realized_output_usd), waited for on-chain rather than taken from the quote."
-  - "Ticket sizes: $3 and $30. Small on purpose, since this measures the quote-to-settlement gap, not large-ticket depth."
+  - "Realized fee: ticket minus the value that landed on the destination chain, in basis points of the ticket (bridge_exec_realized_fee_bps), so $3 and $30 executions compare. Published only for settled executions whose destination balance was read on-chain; a leg whose credit was not visible within 30 s publishes nothing and is counted in bridge_realized_fallback_total."
+  - "Ticket sizes: $3 daily and $30 on Monday and Thursday, pooled. Small on purpose, since this measures the quote-to-settlement gap, not large-ticket depth. Window: 7 days, about 7 to 9 executions per bridge and corridor."
   - "Region: EU-West only. The execution node runs a single wallet, so a second origin would race the same inventory."
-  - "Cohort: Mobula, Relay and LI.FI, the bridges executed on all three legs. Quote-only providers are covered by the bridge-fee and bridge-quote-latency benchmarks."
+  - "Cohort: Mobula, Relay, LI.FI and Near Intents, the bridges executed on all three legs. Quote-only providers are covered by the bridge-fee and bridge-quote-latency benchmarks."
 
 findings:
-  - "{{name:mobula}} settles within {{p50:mobula}} of its quote (median, 24 h) on executed USDC transfers."
+  - "{{name:mobula}} settles within {{p50:mobula}} of its quote (median, 7 d) on executed USDC transfers."
   - "{{name:relay}} settles within {{p50:relay}} of its quote."
   - "{{name:lifi}} settles within {{p50:lifi}} of its quote."
+  - "{{name:near-intents}} settles within {{p50:near-intents}} of its quote."
 
 faq:
   - q: "How is this different from the bridge fee benchmark?"
     a: "The fee benchmark measures the cost a quote advertises. This one measures the cost that actually settled on-chain and subtracts what the quote promised, so it captures the gap a user only discovers after they commit. A bridge can quote cheap and settle worse, and only an execution benchmark surfaces that."
   - q: "Why can the number be negative?"
-    a: "Because a bridge can settle better than it quoted. Slippage vs quote is realized fee minus quoted fee, so a solver that fills above the quoted output makes the figure negative. Near zero means the quote was honest; positive means the execution cost more than promised."
+    a: "Because a bridge can settle better than it quoted. Slippage vs quote is realized fee minus quoted fee, both defined against the ticket (ticket minus landed, ticket minus quoted output), so a solver that fills above the quoted output makes the figure negative. Near zero means the quote was honest; positive means the execution cost more than promised."
   - q: "Does this include the gas I pay?"
     a: "The headline is the bridge's own quote-to-settlement gap. Our own execution gas (approve plus deposit) is reported separately in the all-in gas panel, measured as the real native-token balance delta, because it is a cost the user bears on top of the bridge fee."
   - q: "Where does this run from?"
@@ -71,26 +73,19 @@ source: https://github.com/ChainBench/OpenChainBench/tree/main/harnesses/bridge-
 
 metric_panels:
   - id: fee
-    label: Realized fee
-    metric: avg by (bridge) (ocb:bridge_fees_usd:avg_7d{region="eu-west"})
+    label: Realized fee (bps of ticket)
+    metric: avg by (bridge) (ocb:bridge_exec_realized_fee_bps:avg_7d{region="eu-west"})
     label_key: bridge
-    unit: usd
+    unit: bps
     higher_is_better: false
-    description: "Amount sent minus amount actually received on the destination chain, in USD, averaged over the last 7 days of executions. The fee a user really paid, read on-chain rather than from the quote."
+    description: "Ticket minus the value that actually landed on the destination chain, in basis points of the ticket so the $3 and $30 executions compare, averaged over the last 7 days of settled executions whose destination balance was read on-chain."
   - id: gas
     label: All-in gas we paid
     metric: avg by (bridge) (ocb:bridge_exec_gas_usd:avg_7d{region="eu-west"})
     label_key: bridge
     unit: usd
     higher_is_better: false
     description: "Our own on-chain gas (approve plus deposit), measured as the source-chain native balance delta. The cost the user bears on top of the bridge fee."
-  - id: realized
-    label: Realized output on-chain
-    metric: avg by (bridge) (ocb:bridge_realized_output_usd:avg_7d{region="eu-west"})
-    label_key: bridge
-    unit: usd
-    higher_is_better: true
-    description: "The USD value that actually landed on the destination chain, waited for on-chain rather than read from the quote."
 
 prometheus:
   window: 7d

@@ -41,9 +41,11 @@ type ExecutionResult struct {
 	ExecutionLatencyMs  int64 // Settlement latency: destination block timestamp minus source block timestamp (on-chain); wall clock when a hash is missing
 	ObservedLatencyMs   int64 // Wall clock from broadcast to the poll that saw the terminal status (the pre-2026-09-19 figure), kept for the audit trail
 	DestTxHash          string
-	LatencyMethod       string // "watch" (credit observed minus source inclusion observed, ms) | "watch-broadcast" (credit minus broadcast) | "blocks" (block timestamp delta) | "poll"
-	OnchainBlockDeltaMs int64  // destination block timestamp minus source block timestamp, -1 when unknown
-	E2ELatencyMs        int64  // Time from quote start to funds received
+	LatencyMethod       string  // "watch" (credit observed minus source inclusion observed, ms) | "watch-broadcast" (credit minus broadcast) | "blocks" (block timestamp delta) | "poll"
+	OnchainBlockDeltaMs int64   // destination block timestamp minus source block timestamp, -1 when unknown
+	QuotedOutputUSD     float64 // what the quote said would land (USD), before any on-chain read
+	RealizedOnChain     bool    // OutputUSD and ActualFeeUSD come from the destination balance delta, not from the quote
+	E2ELatencyMs        int64   // Time from quote start to funds received
 	Success             bool
 	Reverted            bool
 	Refunded            bool // subset of Reverted: provider returned capital (status "refunded")
@@ -438,6 +440,17 @@ func (e *Executor) executeOnBridge(bridge string, route TestRoute, amount, amoun
 	// Prometheus correctly classify it (Reverted takes precedence over Success).
 	result.Success = !result.Reverted
 
+	// One definition of the quoted fee for every bridge: ticket minus the
+	// output the quote promised. Provider fee sums mixed in gas paid in the
+	// native token (Mobula: TotalFeeUsd + GasFeeUsd), which the realized fee
+	// (ticket minus landed) can never contain, so their slippage read a
+	// constant negative offset regardless of ticket (-$0.0015 at $3 and at
+	// $30 on Base). Gas we pay ourselves is its own panel.
+	result.QuotedOutputUSD = result.OutputUSD
+	if amountUSD > 0 && result.QuotedOutputUSD > 0 && result.QuotedOutputUSD <= amountUSD {
+		result.QuoteFeeUSD = amountUSD - result.QuotedOutputUSD
+	}
+
 	// Read the destination balance again to compute the REALIZED fill on-chain.
 	// Bridge status "filled" sometimes precedes the destination credit by 1-3
 	// blocks; pollRealizedFill waits up to 30s for the delta to materialise.
@@ -451,6 +464,7 @@ func (e *Executor) executeOnBridge(bridge string, route TestRoute, amount, amoun
 			realizedUSD := realizedToken * destinationUSDPerToken(route)
 			log.Printf("    💰 Realized fill on-chain: %.6f tokens = $%.4f (quote projected $%.4f)", realizedToken, realizedUSD, result.OutputUSD)
 			result.OutputUSD = realizedUSD
+			result.RealizedOnChain = true
 			// Recompute fees from realized: amount sent - amount received
 			realFees := amountUSD - realizedUSD
 			if realFees < 0 {
@@ -1015,20 +1029,31 @@ func (e *Executor) recordExecutionMetrics(result *ExecutionResult) {
 	}
 
 	// Record fees + the new execution-cost metrics
-	pulse(bridgeFeesUSD, labels, result.ActualFeeUSD)
-	if result.AmountUSD > 0 {
-		pulse(bridgeFeesPercent, labels, (result.ActualFeeUSD/result.AmountUSD)*100)
-	}
-	if result.OutputUSD > 0 {
-		pulse(bridgeRealizedOutputUSD, labels, result.OutputUSD)
+	// Realized cost metrics come from the destination balance delta only.
+	// A failed leg, a 30 s destination poll miss or a pre-balance read
+	// failure leaves OutputUSD at the quote's projection: published as
+	// "realized", that was a quote number wearing an on-chain label (Relay
+	// $30 Base: one errored leg published $29.97 of "landed" value).
+	if result.Success && result.RealizedOnChain {
+		pulse(bridgeFeesUSD, labels, result.ActualFeeUSD)
+		pulse(bridgeExecRealizedFeeUSD, labels, result.ActualFeeUSD)
+		if result.AmountUSD > 0 {
+			pulse(bridgeFeesPercent, labels, (result.ActualFeeUSD/result.AmountUSD)*100)
+			pulse(bridgeExecRealizedFeeBps, labels, (result.ActualFeeUSD/result.AmountUSD)*10000)
+		}
+		if result.OutputUSD > 0 {
+			pulse(bridgeRealizedOutputUSD, labels, result.OutputUSD)
+		}
+	} else if result.Success {
+		bridgeRealizedFallback.WithLabelValues(result.Bridge, result.FromChain, result.ToChain, e.region).Inc()
 	}
 	// Execution slippage vs quote = realized fee - quote-projected fee. Only on
 	// a real fill: on a revert / refund / pre-broadcast failure there is no
 	// realized fee (ActualFeeUSD stays 0 while QuoteFeeUSD was set), so recording
 	// it would inject spurious 0 / negative samples into the realized-cost bench.
 	// Slippage is only meaningful against a quoted fee; Near Intents used to
 	// publish its whole realized fee here (QuoteFeeUSD was never set).
-	if result.Success && result.QuoteFeeUSD > 0 {
+	if result.Success && result.RealizedOnChain && result.QuoteFeeUSD > 0 {
 		pulse(bridgeQuoteSlippageUSD, labels, result.ActualFeeUSD-result.QuoteFeeUSD)
 	}
 	if result.ExecGasUSD > 0 {

@@ -143,6 +143,22 @@ var (
 		Help: "Settled executions whose published latency did not come from the two-ended watch, by method (watch-broadcast: source inclusion missed; blocks: block-timestamp delta; poll: wall clock to the status poll)",
 	}, []string{"bridge", "from_chain", "to_chain", "region", "method"})
 
+	// Execution-only realized fee (bridge_fees_usd is also written by the
+	// quote loop at $5/$50/$300 and off-triangle routes, so a panel reading
+	// it averaged quotes into "realized"). Pulses.
+	bridgeExecRealizedFeeUSD = promauto.NewGaugeVec(prometheus.GaugeOpts{
+		Name: "bridge_exec_realized_fee_usd",
+		Help: "Realized fee of one settled execution in USD: ticket minus the value that landed on the destination, read from the balance delta (pulse)",
+	}, []string{"bridge", "from_chain", "to_chain", "from_token", "to_token", "amount_usd", "region", "chain"})
+	bridgeExecRealizedFeeBps = promauto.NewGaugeVec(prometheus.GaugeOpts{
+		Name: "bridge_exec_realized_fee_bps",
+		Help: "Realized fee of one settled execution in basis points of the ticket (comparable across $3 and $30) (pulse)",
+	}, []string{"bridge", "from_chain", "to_chain", "from_token", "to_token", "amount_usd", "region", "chain"})
+	bridgeRealizedFallback = promauto.NewCounterVec(prometheus.CounterOpts{
+		Name: "bridge_realized_fallback_total",
+		Help: "Settled executions whose destination balance could not be read (pre-balance read failed or the credit was not visible within 30 s): no realized fee, output or slippage published for them",
+	}, []string{"bridge", "from_chain", "to_chain", "region"})
+
 	// Persistent: unix time of the last completed execution per bridge and
 	// region, the site's freshness source (prometheus.freshness_timestamp_metric).
 	bridgeLastExecutionTs = promauto.NewGaugeVec(prometheus.GaugeOpts{

@@ -33,7 +33,9 @@ groups:
         expr: avg_over_time(bridge_exec_gas_usd[7d])
       - record: ocb:bridge_realized_output_usd:avg_7d
         expr: avg_over_time(bridge_realized_output_usd[7d])
-      - record: ocb:bridge_fees_usd:avg_7d
-        expr: avg_over_time(bridge_fees_usd[7d])
+      - record: ocb:bridge_exec_realized_fee_usd:avg_7d
+        expr: avg_over_time(bridge_exec_realized_fee_usd[7d])
+      - record: ocb:bridge_exec_realized_fee_bps:avg_7d
+        expr: avg_over_time(bridge_exec_realized_fee_bps[7d])
       - record: ocb:bridge_exec_onchain_ms:mean_7d
         expr: avg_over_time(bridge_exec_onchain_ms[7d])
@@ -549,7 +549,11 @@ export default async function BenchmarkPage({
         )}
         {!isDraft && !insufficient && (
           <span className="ml-auto">
-            <LiveIndicator lastRunAt={benchmark.lastRunAt} slug={benchmark.slug} />
+            <LiveIndicator
+              lastRunAt={benchmark.lastRunAt}
+              slug={benchmark.slug}
+              staleAfterSec={Math.max(300, benchmark.expectedFreshnessSec ?? 300)}
+            />
           </span>
         )}
       </div>

@@ -122,7 +122,7 @@ export function BenchInfobox({ benchmark }: { benchmark: Benchmark }) {
             ) : null}
           </InfoPair>
           {top && (
-            <InfoPair label="Leader (24h)">
+            <InfoPair label={`Leader (${benchmark.window ?? "24h"})`}>
               <span className="font-medium text-ink">{top.name}</span>
               <span className="text-ink-faint">
                 {" "}

@@ -19,7 +19,12 @@ import { useEffect, useState } from "react";
 export function LiveIndicator({
   lastRunAt,
   slug,
+  staleAfterSec = 300,
 }: {
+  /** Seconds after which the dot turns "Stale". Benches whose data moves
+   *  once a day (bridge executions) pass their expected freshness; the
+   *  default is the 5-minute ISR rhythm. */
+  staleAfterSec?: number;
   lastRunAt: string;
   slug?: string;
 }) {
@@ -72,7 +77,7 @@ export function LiveIndicator({
   }, [slug]);
 
   const ageSec = Math.max(0, Math.floor((now - new Date(canonical).getTime()) / 1000));
-  const stale = ageSec > 300;
+  const stale = ageSec > staleAfterSec;
 
   return (
     <span className="inline-flex items-center gap-2 normal-case tracking-normal text-xs tabular text-ink-muted">

@@ -19,11 +19,12 @@ export function StaticLedger({ benchmark }: { benchmark: Benchmark }) {
   if (rows.length === 0) return null;
   const chain = rpcChainLabel(benchmark);
   const pausedOn = isStaleBench(benchmark) && benchmark.lastRunAt ? benchmark.lastRunAt.slice(0, 10) : null;
+  const win = benchmark.window ?? "24h";
   const heading = chain && pausedOn
     ? `Results: measurement paused since ${pausedOn}, last ranking of ${rows.length} free public ${chain} RPC endpoint${rows.length === 1 ? "" : "s"}`
     : chain
     ? `Results: ${rows.length} free public ${chain} RPC endpoint${rows.length === 1 ? "" : "s"} ranked by p50 latency (24h, 3 regions)`
-    : `Results: ${rows.length} providers ranked by ${benchmark.metric} (p50, 24h)`;
+    : `Results: ${rows.length} providers ranked by ${benchmark.metric} (p50, ${win})`;
   const showTail = benchmark.unit === "ms" || benchmark.unit === "s";
   return (
     <section className="mt-8" aria-labelledby="results">
@@ -33,7 +34,7 @@ export function StaticLedger({ benchmark }: { benchmark: Benchmark }) {
       <div className="overflow-x-auto">
         <table className="w-full border-collapse text-sm">
           <caption className="sr-only">
-            {benchmark.title}: {benchmark.metric} per provider, last 24 hours.
+            {benchmark.title}: {benchmark.metric} per provider, last {win === "24h" ? "24 hours" : win}.
           </caption>
           <thead>
             <tr className="border-y-2 border-ink text-left">