diff --git a/scripts/check-cli-startup-memory.mjs b/scripts/check-cli-startup-memory.mjs
index bc88ce97c23c..8b547ca7aafd 100644
--- a/scripts/check-cli-startup-memory.mjs
+++ b/scripts/check-cli-startup-memory.mjs
@@ -99,9 +99,9 @@ function resolveDefaultLimitsMb(platform = process.platform) {
     // Plugin discovery is heavier than help, but must stay below the doctor/channel
     // runtime graph that an empty metadata-only invocation must not import.
     pluginsList: platform === "darwin" ? 500 : 400,
-    // Node 24 status startup sits near 400 MB; retain regression signal without
-    // failing on sub-megabyte runner RSS variance.
-    statusJson: 425,
+    // Node 24 status startup reaches ~430 MB on current Linux runner images;
+    // retain useful regression headroom without failing on allocator variance.
+    statusJson: 450,
     gatewayStatus: 500,
   };
 }
diff --git a/test/scripts/check-cli-startup-memory.test.ts b/test/scripts/check-cli-startup-memory.test.ts
index 87adc9ce228f..258fef1bb290 100644
--- a/test/scripts/check-cli-startup-memory.test.ts
+++ b/test/scripts/check-cli-startup-memory.test.ts
@@ -92,7 +92,7 @@ describe("check-cli-startup-memory", () => {
   });
 
   it("keeps status startup headroom above Linux runner RSS variance", () => {
-    expect(testing.resolveDefaultLimitsMb("linux").statusJson).toBe(425);
+    expect(testing.resolveDefaultLimitsMb("linux").statusJson).toBe(450);
   });
 
   it("uses the median of three cold-start RSS samples", () => {
@@ -115,10 +115,11 @@ describe("check-cli-startup-memory", () => {
       return;
     }
 
-    const helpSamplesMb = [120, 110, 80];
+    const helpLimitMb = testing.resolveDefaultLimitsMb(process.platform).help;
+    const helpSamplesMb = [helpLimitMb + 20, helpLimitMb + 10, 80];
 
     expect(() => runStartupMemoryCheckWithHelpSamples(helpSamplesMb)).toThrow(
-      "--help median max RSS 110.0 MB exceeded 100 MB",
+      `--help median max RSS ${(helpLimitMb + 10).toFixed(1)} MB exceeded ${helpLimitMb} MB`,
     );
   });