diff --git a/restoration/source410/BT/MECH.CPP b/restoration/source410/BT/MECH.CPP
index 49939d78..b7dc6081 100644
--- a/restoration/source410/BT/MECH.CPP
+++ b/restoration/source410/BT/MECH.CPP
@@ -546,6 +546,15 @@ Mech::Mech(
 				if (model->maxAcceleration > 1.0f && model->maxAcceleration < 500.0f)
 				{
 					maxBodyAcceleration = model->maxAcceleration;
+					//
+					// The gait's slew rates ARE the authored acceleration
+					// (binary ctor: +0x344/+0x5b0/+0x5b8 all take rec+0x44).
+					// This is what makes the stride-driven speed model and
+					// the old acceleration model agree in feel: both slew at
+					// maxAcceleration toward the demand.
+					//
+					forwardCycleRate = model->maxAcceleration;
+					gimpCycleRate    = model->maxAcceleration;
 				}
 				if (model->throttleAdjustment > 0.05f && model->throttleAdjustment < 20.0f)
 				{
@@ -1221,29 +1230,39 @@ void
 	// runs as a pure stride measurement (move_joints 0) so the two never
 	// fight over the same joints.
 	//
-	// STAGED SEAM: the binary's caller is IntegrateMotion (mech4 @004ab1c8),
-	// which also USES the body distance as the mech's forward step and picks
-	// the airborne flavours by movement mode.  Until mech4 lands, the body
-	// distance is measured-and-dropped and the acceleration model below
-	// still carries the hull; the two speeds agree because both track the
-	// same demand.
+	// The body channel's measured stride IS the mech's speed (the binary:
+	// IntegrateMotion @004ab1c8 sets local velocity z = -distance/dt).  The
+	// gait slews its cycle speed at forwardCycleRate == the authored
+	// maxAcceleration, so this replaces the earlier explicit acceleration
+	// model at the same rate -- but the speed now rises THROUGH the gait:
+	// zero while standing, the walk band during the walk cycle, stepping to
+	// the run band at a clip boundary.
+	//
+	// STILL STAGED from IntegrateMotion: the airborne flavour pick, the
+	// dead-reckon latency fold, and the turn-in-place dispatcher (mech4).
 	//-----------------------------------------------------------------------
 	//
 	AdvanceLegAnimation(time_slice);
-	AdvanceBodyAnimation(time_slice, 0);
-
-	//
-	//-----------------------------------------------------------------------
-	// Accelerate the actual body speed toward the demand (bounded per frame by
-	// the mech's max acceleration).
-	//-----------------------------------------------------------------------
-	//
 	{
-		Scalar dv      = bodyTargetSpeed - currentBodySpeed;
-		Scalar maxStep = maxBodyAcceleration * time_slice;
-		if (dv >  maxStep) dv =  maxStep;
-		if (dv < -maxStep) dv = -maxStep;
-		currentBodySpeed += dv;
+		Scalar stride = AdvanceBodyAnimation(time_slice, 0);
+
+		if (animationClips[5] != ResourceDescription::NullResourceID)
+		{
+			currentBodySpeed = stride / time_slice;
+		}
+		else
+		{
+			//
+			// [T3 bring-up] A model with NO gait clips would otherwise never
+			// move.  Keep the old acceleration model for those until every
+			// fleet mech's clip set is verified.
+			//
+			Scalar dv      = bodyTargetSpeed - currentBodySpeed;
+			Scalar maxStep = maxBodyAcceleration * time_slice;
+			if (dv >  maxStep) dv =  maxStep;
+			if (dv < -maxStep) dv = -maxStep;
+			currentBodySpeed += dv;
+		}
 	}
 
 	//
diff --git a/restoration/source410/BT/MECH2.NOTES.md b/restoration/source410/BT/MECH2.NOTES.md
index b8208b14..72946136 100644
--- a/restoration/source410/BT/MECH2.NOTES.md
+++ b/restoration/source410/BT/MECH2.NOTES.md
@@ -289,3 +289,35 @@ running correctly. No fault, mission drove itself clean.
   writers is open work (model resource? mech3/mech4?).
 * `ForceUpdate(8)` (the leg-state update-record request, binary @004a4c54
   inline) is a STAGED no-op — our replication emitter is not reconstructed.
+
+## The stride became the speed (2026-08-03)
+
+The STAGED seam in `Simulate` is half-closed: the body channel's measured
+distance is now the mech's speed (`currentBodySpeed = stride / dt`), which is
+the binary's own relationship (`IntegrateMotion` @004ab1c8 sets local velocity
+z = −distance/dt). The explicit acceleration model survives only as a [T3]
+fallback for a model with no gait clips at all.
+
+Two findings made this safe rather than speculative:
+
+**The gait slew rate IS the authored acceleration.** The binary's ctor sets
+`forwardCycleRate`, `gimpCycleRate` and `groundCycleRate` all from the model's
+maxAcceleration (rec+0x44; 30 for the Mad Cat), and `airborneCycleRate` from
+superStopAcceleration. So the stride-driven speed slews at exactly the rate
+the old model used — feel preserved, source authentic. Our ctor now sets
+`forwardCycleRate`/`gimpCycleRate` from the same guarded model read. (This
+also retires the ctor default of 1.0, which would have taken ~27 s to reach
+speed.)
+
+**Live curve, arena run:** first sample 7.32 (stepping off through the
+stand→walk clip), then 22–30 oscillating around the demanded 26.9 — the
+per-stride speed variation of a real gait, where the old model pinned
+26.9022 flat. Cadence in the hull motion is the authentic behaviour.
+
+Still staged from IntegrateMotion: the airborne flavour pick
+(`movementMode 3/4 && +0x580`), the dead-reckon latency fold (+0x778/+0x77c),
+the local-velocity/orientation integration proper, and the turn-in-place
+dispatcher. One correction queued by the raw decomp: `+0x598` is written by a
+VECTOR op in IntegrateMotion's replicant branch (FUN_00408644 on floats), so
+`motionEventName` is likely a Point3D, not a CString — revisit when that
+branch is reconstructed.