diff --git a/CHANGELOG.md b/CHANGELOG.md
index 3a32ec07d9..6b471ad001 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -10,6 +10,7 @@
 
 ### Changes
 
+- Renamed `make update-note-costs` to `make update-generated`, which regenerates the note cost tables and `bench-tx.json` together and is run only when the cost drift guard fails ([#3727](https://github.com/0xMiden/protocol/pull/3727)).
 - Keyed `bench-tx.json`'s per-note cycle counts by note label instead of note commitment, and added `total_cycles` to each scenario ([#3726](https://github.com/0xMiden/protocol/pull/3726)).
 - [BREAKING] Refactored `AccountVaultDelta` to track generic assets. `FungibleAssetDelta`, `NonFungibleAssetDelta` and `NonFungibleDeltaAction` were removed ([3485](https://github.com/0xMiden/protocol/pull/3485)).
 - [BREAKING] Moved the internal shared helpers of `miden::protocol::input_note`, `miden::protocol::active_note`, and the note memory-write helpers into private `input_note_internal` and `note_internal` modules ([#3501](https://github.com/0xMiden/protocol/pull/3501)).
diff --git a/Makefile b/Makefile
index e07ccbbf76..66ee899730 100644
--- a/Makefile
+++ b/Makefile
@@ -175,9 +175,9 @@ bench-tx: ## Run transaction benchmarks
 bench-note-checker: ## Run note checker benchmarks
 	cargo bench --bin bench-note-checker --bench benches
 
-.PHONY: update-note-costs
-update-note-costs: ## Regenerate bench-tx.json and the checked-in note consumption cost tables
-	cargo run --bin bench-transaction --features concurrent -- update-note-costs
+.PHONY: update-generated
+update-generated: ## Regenerate bench-tx.json and the checked-in note consumption cost tables
+	cargo run --release --bin bench-transaction --features concurrent -- update-generated
 	cargo +nightly fmt -p miden-standards -p miden-agglayer
 
 # --- installing ----------------------------------------------------------------------------------
diff --git a/bin/bench-transaction/README.md b/bin/bench-transaction/README.md
index 6401573e08..17b098a940 100644
--- a/bin/bench-transaction/README.md
+++ b/bin/bench-transaction/README.md
@@ -34,13 +34,33 @@ The network-account auth procedure collects sponsored fees and answers sponsorsh
 
 The network-account scenarios feed two checked-in, generated cost tables: `crates/miden-standards/src/note/costs/table.rs` and `crates/miden-agglayer/src/costs/table.rs`. Each table entry is the note's consumption cost in VM cycles - the total cycle count of the canonical network-account transaction consuming it, taken as the maximum across the note's benchmarked execution paths. The values are estimates, not guaranteed worst cases - see the caveats in `miden_standards::note::costs` (e.g. asset counts are benchmarked at the planned, not current, protocol maximum).
 
-Regenerate the tables (and `bench-tx.json`) with:
+### Regenerating the Generated Artifacts
+
+There are two of them - the cost tables above and `bench-tx.json` - and they are regenerated
+together, only when the drift guard asks for it:
 
 ```bash
-make update-note-costs
+make update-generated
 ```
 
-Freshness is enforced in CI: the `checked_in_note_costs_match_executed_scenarios` snapshot test in `src/note_costs.rs` re-executes every priced scenario during the regular test run and fails when a measured cost drifts more than 5% from its checked-in constant, or when a note's declared `created_notes` do not match what its scenarios actually create. It walks `PricedNote::all` - so all priced notes are covered. Drift within the tolerance (from unrelated changes landing on the base branch) is absorbed without regeneration - fee-wise this is safe, since the fee is logarithmic in cycles and the pricing safety margin dwarfs it. A PR that meaningfully changes cycle counts must run `make update-note-costs` and commit the updated tables - which doubles as review signal, since cost regressions show up as table diffs.
+The guard is the `checked_in_note_costs_match_executed_scenarios` test in `src/note_costs.rs`. It
+re-executes every priced scenario during the regular test run and fails when a measured cost drifts
+more than 5% from its checked-in constant, or when a note's declared `created_notes` do not match
+what its scenarios actually create. It walks `PricedNote::all` - so all priced notes are covered.
+
+Until that test fails, a PR regenerates nothing, including a PR that changes cycle counts.
+Sub-threshold drift is deliberately absorbed. Fee-wise that is safe: the fee is logarithmic in
+cycles and the pricing safety margin dwarfs the tolerated drift. It is also the cheaper trade for
+review - regenerating on every cost-affecting change buries a couple of hundred lines of real diff
+under a thousand lines of generated churn, which is what the tables and the snapshot produce
+between them.
+
+When the guard does fail, run `make update-generated` and commit both artifacts in the same change.
+Keeping them in lockstep is why there is one target rather than two.
+
+`bench-tx.json` is refreshed only as part of that regeneration, so between refreshes it is not
+expected to match the current tree, and nothing enforces that it does. It is an observational
+snapshot.
 
 ### Benchmark Groups
 
diff --git a/bin/bench-transaction/src/main.rs b/bin/bench-transaction/src/main.rs
index abc3eae897..8b7b6ef51f 100644
--- a/bin/bench-transaction/src/main.rs
+++ b/bin/bench-transaction/src/main.rs
@@ -30,10 +30,10 @@ async fn run_scenario(
 
 #[tokio::main(flavor = "current_thread")]
 async fn main() -> Result<()> {
-    let update_costs = match std::env::args().nth(1).as_deref() {
+    let update_cost_tables = match std::env::args().nth(1).as_deref() {
         None => false,
-        Some("update-note-costs") => true,
-        Some(other) => anyhow::bail!("unknown argument `{other}`; expected `update-note-costs`"),
+        Some("update-generated") => true,
+        Some(other) => anyhow::bail!("unknown argument `{other}`; expected `update-generated`"),
     };
 
     // create a template file for benchmark results
@@ -52,7 +52,7 @@ async fn main() -> Result<()> {
     // store benchmark results in the JSON file
     write_bench_results_to_json(path, benchmark_results)?;
 
-    if update_costs {
+    if update_cost_tables {
         bench_transaction::note_costs::update_cost_tables(&measured_cycles)?;
     }
 
diff --git a/bin/bench-transaction/src/note_costs.rs b/bin/bench-transaction/src/note_costs.rs
index 95c8ccdfcb..57eb396c72 100644
--- a/bin/bench-transaction/src/note_costs.rs
+++ b/bin/bench-transaction/src/note_costs.rs
@@ -17,7 +17,7 @@ use miden_standards::note::costs::NoteCost;
 use crate::cycle_counting_benchmarks::ExecutionBenchmark;
 
 /// Header line of the generated table files.
-const GENERATED_HEADER: &str = "// This file is @generated by `make update-note-costs`. Do not edit by hand.\n\
+const GENERATED_HEADER: &str = "// This file is @generated by `make update-generated`. Do not edit by hand.\n\
 // Values are maxima across the benchmarked paths; see `miden_standards::note::costs` for the\n\
 // caveats on what they do and do not cover.\n";
 
@@ -478,8 +478,8 @@ mod tests {
         if !stale_costs.is_empty() {
             failures.push(format!(
                 "cost table stale for {} note(s), each more than {DRIFT_TOLERANCE_PERCENT}% from \
-                 its checked-in constant: {}. Run `make update-note-costs` and commit the updated \
-                 tables",
+                 its checked-in constant: {}. Run `make update-generated` and commit both \
+                 regenerated artifacts - the cost tables and bench-tx.json - together",
                 stale_costs.len(),
                 stale_costs.join("; "),
             ));
diff --git a/crates/miden-agglayer/src/costs/mod.rs b/crates/miden-agglayer/src/costs/mod.rs
index 0794076a51..61813ae2e8 100644
--- a/crates/miden-agglayer/src/costs/mod.rs
+++ b/crates/miden-agglayer/src/costs/mod.rs
@@ -7,7 +7,7 @@
 //! The `NetworkNotePricer` in `miden-tx` turns cycle costs into fees, resolving the agglayer
 //! notes through [`AgglayerNote::note_cost`](crate::AgglayerNote::note_cost).
 //!
-//! The table is regenerated with `make update-note-costs`; a snapshot test in
+//! The table is regenerated with `make update-generated`; a snapshot test in
 //! `bench-transaction` fails CI when a checked-in value drifts more than 5% from the measured
 //! one (small drift from unrelated changes is tolerated - the pricing safety margin dwarfs
 //! it).
diff --git a/crates/miden-agglayer/src/costs/table.rs b/crates/miden-agglayer/src/costs/table.rs
index 535a1760aa..673093af87 100644
--- a/crates/miden-agglayer/src/costs/table.rs
+++ b/crates/miden-agglayer/src/costs/table.rs
@@ -1,4 +1,4 @@
-// This file is @generated by `make update-note-costs`. Do not edit by hand.
+// This file is @generated by `make update-generated`. Do not edit by hand.
 // Values are maxima across the benchmarked paths; see `miden_standards::note::costs` for the
 // caveats on what they do and do not cover.
 
diff --git a/crates/miden-standards/src/note/costs/mod.rs b/crates/miden-standards/src/note/costs/mod.rs
index 0573300119..f846247ad9 100644
--- a/crates/miden-standards/src/note/costs/mod.rs
+++ b/crates/miden-standards/src/note/costs/mod.rs
@@ -20,7 +20,7 @@
 //! Terminology: a note's *cost* is its measured cycle count; its *price* is the fee derived
 //! from that cost (and from the costs of the notes its consumption creates).
 //!
-//! The table is regenerated with `make update-note-costs`; a snapshot test in
+//! The table is regenerated with `make update-generated`; a snapshot test in
 //! `bench-transaction` fails CI when a checked-in value drifts more than 5% from the measured
 //! one (small drift from unrelated changes is tolerated - the pricing safety margin dwarfs
 //! it).
diff --git a/crates/miden-standards/src/note/costs/table.rs b/crates/miden-standards/src/note/costs/table.rs
index 8b925d0264..b804976bdb 100644
--- a/crates/miden-standards/src/note/costs/table.rs
+++ b/crates/miden-standards/src/note/costs/table.rs
@@ -1,4 +1,4 @@
-// This file is @generated by `make update-note-costs`. Do not edit by hand.
+// This file is @generated by `make update-generated`. Do not edit by hand.
 // Values are maxima across the benchmarked paths; see `miden_standards::note::costs` for the
 // caveats on what they do and do not cover.