diff --git a/ci/source-shape-test-budget.json b/ci/source-shape-test-budget.json
index 5eb801c3dfc..a93ce1a21e1 100644
--- a/ci/source-shape-test-budget.json
+++ b/ci/source-shape-test-budget.json
@@ -76,6 +76,11 @@
       "test": "restores exact locked posture after root-separated repair and later failure (#7033)",
       "category": "security"
     },
+    {
+      "file": "test/muse-glimmer-vllm-image-provenance.test.ts",
+      "test": "binds the checked-in provenance to the selected runtime",
+      "category": "security"
+    },
     {
       "file": "test/muse-glimmer-vllm-image-provenance.test.ts",
       "test": "rejects %s",
diff --git a/docs/inference/set-up-vllm.mdx b/docs/inference/set-up-vllm.mdx
index f155ccbb4cd..25b2ffa46f1 100644
--- a/docs/inference/set-up-vllm.mdx
+++ b/docs/inference/set-up-vllm.mdx
@@ -151,8 +151,9 @@ Managed profiles and model-specific recipes use immutable image digests:
 
 - DGX Spark and DGX Station models without a model-specific runtime use the `linux/arm64` digest `sha256:9204569b17ee4c0eff75194b8e6e458479c8aee18953b5ab9cf359fcdac659e2` with a compressed layer size of `9.60 GB` under `nvcr.io/nvidia/vllm:26.05.post1-py3`.
   N1x uses the same image.
-- The DGX Spark Muse Glimmer recipe uses the `linux/arm64` post-merge vLLM nightly digest `sha256:677afd5bf3b4bb9881f91e107af7098f8410726b4c05b25cb4a815900b398204` with a compressed layer size of `9.70 GB` under `vllm/vllm-openai`.
-  The image was built from vLLM commit `ac7509e2b1db40fec2f03dde1ed4e9dfdc2338c9`, which contains merged Muse Glimmer support commit `6adad08767583f52eb4d2122111af0bf638ed5e6`.
+- The DGX Spark Muse Glimmer recipe uses the `linux/arm64` post-fix vLLM nightly digest `sha256:b0e84e5f2b00a7268e4fdda332790ebd4bfb166b64757e166914753afaeee965` with a compressed layer size of `9.71 GB` under `vllm/vllm-openai`.
+  The image was built from vLLM commit `5a4c8d99242e9e069b604d0e9b969e77f7dd501d`, which contains merged Muse Glimmer support commit `6adad08767583f52eb4d2122111af0bf638ed5e6` and the vLLM dependency fix from commit `90984ddbed27a09409506d6d6c0eea87f54b04b5`.
+  The image includes `huggingface_hub` `1.28.0`, which preserves the pinned model revision when vLLM transfers configuration to the engine process.
   It pins Hugging Face revision `d35cb79050f419c457611b1cee5c5d15b176f285` for the approximately `25.45 GB` model download.
 - The DGX Spark Nemotron 3.5 Lightning recipe uses the `linux/arm64` digest `sha256:3af90144a0926e5c5fe46ee16e5201e763dd854538b9d7ce433755f11dadaf78` with a compressed layer size of approximately `12.69 GB` under `vllm/vllm-openai`.
   It pins Hugging Face revision `0dcd680e5585c791728c83342b311d0a0026dbeb` for the approximately `21.56 GB` model download.
diff --git a/internal/security-reviews/muse-glimmer-vllm-image-provenance-v1.json b/internal/security-reviews/muse-glimmer-vllm-image-provenance-v1.json
index e28d54fdde4..071bf214c8b 100644
--- a/internal/security-reviews/muse-glimmer-vllm-image-provenance-v1.json
+++ b/internal/security-reviews/muse-glimmer-vllm-image-provenance-v1.json
@@ -11,51 +11,64 @@
     "repository": "vllm/vllm-openai"
   },
   "image": {
-    "reference": "vllm/vllm-openai@sha256:677afd5bf3b4bb9881f91e107af7098f8410726b4c05b25cb4a815900b398204",
-    "manifestDigest": "sha256:677afd5bf3b4bb9881f91e107af7098f8410726b4c05b25cb4a815900b398204",
+    "reference": "vllm/vllm-openai@sha256:b0e84e5f2b00a7268e4fdda332790ebd4bfb166b64757e166914753afaeee965",
+    "manifestDigest": "sha256:b0e84e5f2b00a7268e4fdda332790ebd4bfb166b64757e166914753afaeee965",
     "manifestMediaType": "application/vnd.docker.distribution.manifest.v2+json",
-    "manifestUrl": "https://registry-1.docker.io/v2/vllm/vllm-openai/manifests/sha256:677afd5bf3b4bb9881f91e107af7098f8410726b4c05b25cb4a815900b398204",
-    "configDigest": "sha256:c3f199e54a26d2d7a9a41115cd07ce9d90a6488c5a4e75b17129e1006ce533fd",
+    "manifestUrl": "https://registry-1.docker.io/v2/vllm/vllm-openai/manifests/sha256:b0e84e5f2b00a7268e4fdda332790ebd4bfb166b64757e166914753afaeee965",
+    "configDigest": "sha256:49d2eb65dc2a8dea24e43c27b226f650481ac97d4ba9c567b6e1ca08bc472303",
     "configMediaType": "application/vnd.docker.container.image.v1+json",
-    "configSizeBytes": 34762,
-    "configUrl": "https://registry-1.docker.io/v2/vllm/vllm-openai/blobs/sha256:c3f199e54a26d2d7a9a41115cd07ce9d90a6488c5a4e75b17129e1006ce533fd",
+    "configSizeBytes": 34738,
+    "configUrl": "https://registry-1.docker.io/v2/vllm/vllm-openai/blobs/sha256:49d2eb65dc2a8dea24e43c27b226f650481ac97d4ba9c567b6e1ca08bc472303",
     "layerCount": 32,
-    "compressedLayerSizeBytes": 9699710136,
-    "createdAt": "2026-08-14T05:33:50.528328374Z",
+    "compressedLayerSizeBytes": 9706339423,
+    "createdAt": "2026-08-19T05:34:00.562836767Z",
     "platform": {
       "os": "linux",
       "architecture": "arm64"
     }
   },
   "build": {
-    "sourceRepository": "https://github.com/vllm-project/vllm",
-    "sourceRevision": "ac7509e2b1db40fec2f03dde1ed4e9dfdc2338c9",
-    "sourceRevisionUrl": "https://github.com/vllm-project/vllm/commit/ac7509e2b1db40fec2f03dde1ed4e9dfdc2338c9",
-    "imageTag": "vllm/vllm-openai:nightly-ac7509e2b1db40fec2f03dde1ed4e9dfdc2338c9",
-    "pipelineId": "019d130e-464e-4ff7-b84b-492992c0c06b",
-    "pipelineUrl": "https://buildkite.com/vllm/release-v2/builds/5174"
+    "sourceRevision": "5a4c8d99242e9e069b604d0e9b969e77f7dd501d",
+    "imageTag": "vllm/vllm-openai:nightly-5a4c8d99242e9e069b604d0e9b969e77f7dd501d",
+    "pipelineId": "019d130e-464e-4ff7-b84b-492992c0c06b"
   },
   "upstreamSupport": {
     "museMergeCommit": "6adad08767583f52eb4d2122111af0bf638ed5e6",
-    "museMergeCommitUrl": "https://github.com/vllm-project/vllm/commit/6adad08767583f52eb4d2122111af0bf638ed5e6",
-    "comparisonUrl": "https://github.com/vllm-project/vllm/compare/6adad08767583f52eb4d2122111af0bf638ed5e6...ac7509e2b1db40fec2f03dde1ed4e9dfdc2338c9",
-    "relationship": "direct-descendant",
-    "aheadBy": 1
+    "relationship": "descendant",
+    "aheadBy": 173,
+    "revisionFixCommit": "90984ddbed27a09409506d6d6c0eea87f54b04b5",
+    "revisionFixRelationship": "descendant",
+    "revisionFixAheadBy": 33
   },
   "reportedLabels": {
-    "ai.vllm.build.commit": "ac7509e2b1db40fec2f03dde1ed4e9dfdc2338c9",
+    "ai.vllm.build.commit": "5a4c8d99242e9e069b604d0e9b969e77f7dd501d",
     "ai.vllm.build.pipeline": "019d130e-464e-4ff7-b84b-492992c0c06b",
-    "ai.vllm.build.url": "https://buildkite.com/vllm/release-v2/builds/5174",
-    "ai.vllm.image.tag": "vllm/vllm-openai:nightly-ac7509e2b1db40fec2f03dde1ed4e9dfdc2338c9",
-    "org.opencontainers.image.revision": "ac7509e2b1db40fec2f03dde1ed4e9dfdc2338c9",
-    "org.opencontainers.image.source": "https://github.com/vllm-project/vllm"
+    "ai.vllm.image.tag": "vllm/vllm-openai:nightly-5a4c8d99242e9e069b604d0e9b969e77f7dd501d",
+    "org.opencontainers.image.revision": "5a4c8d99242e9e069b604d0e9b969e77f7dd501d"
+  },
+  "runtimeDependencies": {
+    "vllmVersion": "0.26.1rc1.dev942+g5a4c8d992",
+    "huggingfaceHubVersion": "1.28.0",
+    "sentencepieceVersion": "0.2.2",
+    "tiktokenVersion": "0.14.0"
+  },
+  "revisionSerialization": {
+    "model": "Inferact/Muse-Glimmer-30B-NVFP4-W4A4",
+    "requestedRevision": "d35cb79050f419c457611b1cee5c5d15b176f285",
+    "resolvedRevisionBeforePickle": "d35cb79050f419c457611b1cee5c5d15b176f285",
+    "resolvedRevisionAfterPickle": "d35cb79050f419c457611b1cee5c5d15b176f285",
+    "preserved": true
   },
   "verification": {
-    "observedAt": "2026-08-14T16:33:50Z",
+    "observedAt": "2026-08-19T22:53:46Z",
     "methods": [
       "docker-buildx-imagetools-inspect-raw",
       "docker-image-inspect",
-      "github-compare-api"
+      "github-compare-api",
+      "container-package-imports",
+      "resolved-revision-pickle-round-trip",
+      "dgx-spark-cold-cache-startup",
+      "vllm-openai-api-validation"
     ],
     "signedProvenanceAttestation": "not-available"
   }
diff --git a/managed-inference/recipes/vllm.muse-glimmer-30b-nvfp4-w4a4.spark-single.v1.yaml b/managed-inference/recipes/vllm.muse-glimmer-30b-nvfp4-w4a4.spark-single.v1.yaml
index 1bd4b5cb3a6..b8a43e91686 100644
--- a/managed-inference/recipes/vllm.muse-glimmer-30b-nvfp4-w4a4.spark-single.v1.yaml
+++ b/managed-inference/recipes/vllm.muse-glimmer-30b-nvfp4-w4a4.spark-single.v1.yaml
@@ -13,8 +13,8 @@ spec:
 
   modelRef: vllm.muse-glimmer-30b-nvfp4-w4a4.v1
   runtime:
-    image: vllm/vllm-openai@sha256:677afd5bf3b4bb9881f91e107af7098f8410726b4c05b25cb4a815900b398204
-    imageDownloadSizeBytes: 9699710136
+    image: vllm/vllm-openai@sha256:b0e84e5f2b00a7268e4fdda332790ebd4bfb166b64757e166914753afaeee965
+    imageDownloadSizeBytes: 9706339423
     minimumComputeCapability: 121
     minimumGpuMemoryBytes: 25447097878
     pullTimeoutSeconds: 3600
diff --git a/src/lib/inference/vllm.test.ts b/src/lib/inference/vllm.test.ts
index 10432a4b39f..c126e3cdda7 100644
--- a/src/lib/inference/vllm.test.ts
+++ b/src/lib/inference/vllm.test.ts
@@ -300,9 +300,9 @@ describe("vLLM profile detection", () => {
     const resolved = resolveVllmModelRuntime(profile!, muse!);
     const runtime = resolved.profile;
     expect(runtime.image).toBe(
-      "vllm/vllm-openai@sha256:677afd5bf3b4bb9881f91e107af7098f8410726b4c05b25cb4a815900b398204",
+      "vllm/vllm-openai@sha256:b0e84e5f2b00a7268e4fdda332790ebd4bfb166b64757e166914753afaeee965",
     );
-    expect(runtime.imageDownloadSizeBytes).toBe(9_699_710_136);
+    expect(runtime.imageDownloadSizeBytes).toBe(9_706_339_423);
     expect(runtime.modelDownloadSizeBytes).toBe(25_447_097_878);
 
     const apiKey = "a".repeat(64);
diff --git a/test/managed-inference-catalog-compiler.test.ts b/test/managed-inference-catalog-compiler.test.ts
index 56e4b2fa0c0..6ff92d461fb 100644
--- a/test/managed-inference-catalog-compiler.test.ts
+++ b/test/managed-inference-catalog-compiler.test.ts
@@ -293,7 +293,7 @@ describe("managed inference YAML profile contract", () => {
       runtime: {
         architecture: "arm64",
         image:
-          "vllm/vllm-openai@sha256:677afd5bf3b4bb9881f91e107af7098f8410726b4c05b25cb4a815900b398204",
+          "vllm/vllm-openai@sha256:b0e84e5f2b00a7268e4fdda332790ebd4bfb166b64757e166914753afaeee965",
       },
       execution: {
         materializerRef: "vllm.host-local/v1",
diff --git a/test/muse-glimmer-vllm-image-provenance.test.ts b/test/muse-glimmer-vllm-image-provenance.test.ts
index dedda9f76b6..1268a625c0a 100644
--- a/test/muse-glimmer-vllm-image-provenance.test.ts
+++ b/test/muse-glimmer-vllm-image-provenance.test.ts
@@ -35,6 +35,26 @@ const recipe = YAML.parse(readFileSync(RECIPE_PATH, "utf8")) as {
 };
 
 describe("Muse Glimmer vLLM image provenance", () => {
+  // source-shape-contract: security -- The checked-in provenance and recipe must resolve to the same reviewed external runtime image and size.
+  it("binds the checked-in provenance to the selected runtime", () => {
+    verifyMuseGlimmerVllmImageProvenance(provenance);
+
+    expect(recipe.spec.runtime).toMatchObject({
+      architecture: "arm64",
+      image: MUSE_GLIMMER_VLLM_IMAGE_REFERENCE,
+      imageDownloadSizeBytes: 9_706_339_423,
+    });
+
+    const profile = detectVllmProfile({ platform: "spark", type: "nvidia" });
+    const model = VLLM_MODELS.find(({ envValue }) => envValue === "muse-glimmer-30b");
+    expect(profile).not.toBeNull();
+    expect(model).toBeDefined();
+    expect(resolveVllmRuntimeProfile(profile!, model!)).toMatchObject({
+      image: MUSE_GLIMMER_VLLM_IMAGE_REFERENCE,
+      imageDownloadSizeBytes: 9_706_339_423,
+    });
+  });
+
   // source-shape-contract: security -- Mutating each trust field proves the reviewed provenance record fails closed before a substituted external runtime can be published.
   it.each([
     ["publisher drift", ["publisher", "namespace"], "attacker"],
@@ -43,6 +63,12 @@ describe("Muse Glimmer vLLM image provenance", () => {
     ["source drift", ["build", "sourceRevision"], "0".repeat(40)],
     ["label drift", ["reportedLabels", "org.opencontainers.image.revision"], "0".repeat(40)],
     ["support ancestry drift", ["upstreamSupport", "relationship"], "unverified"],
+    ["runtime dependency drift", ["runtimeDependencies", "huggingfaceHubVersion"], "1.27.0"],
+    [
+      "revision serialization drift",
+      ["revisionSerialization", "resolvedRevisionAfterPickle"],
+      "main",
+    ],
   ] as const)("rejects %s", (_name, pathParts, replacement) => {
     const changed = structuredClone(provenance) as JsonRecord;
     let target = changed;
diff --git a/test/support/muse-glimmer-vllm-image-provenance-test-support.ts b/test/support/muse-glimmer-vllm-image-provenance-test-support.ts
index fc1ff63c35e..34ec6ee7c9d 100644
--- a/test/support/muse-glimmer-vllm-image-provenance-test-support.ts
+++ b/test/support/muse-glimmer-vllm-image-provenance-test-support.ts
@@ -3,13 +3,14 @@
 
 import assert from "node:assert/strict";
 
-const MANIFEST_DIGEST = "sha256:677afd5bf3b4bb9881f91e107af7098f8410726b4c05b25cb4a815900b398204";
+const MANIFEST_DIGEST = "sha256:b0e84e5f2b00a7268e4fdda332790ebd4bfb166b64757e166914753afaeee965";
 export const MUSE_GLIMMER_VLLM_IMAGE_REFERENCE = `vllm/vllm-openai@${MANIFEST_DIGEST}`;
-const CONFIG_DIGEST = "sha256:c3f199e54a26d2d7a9a41115cd07ce9d90a6488c5a4e75b17129e1006ce533fd";
-const SOURCE_REVISION = "ac7509e2b1db40fec2f03dde1ed4e9dfdc2338c9";
+const CONFIG_DIGEST = "sha256:49d2eb65dc2a8dea24e43c27b226f650481ac97d4ba9c567b6e1ca08bc472303";
+const SOURCE_REVISION = "5a4c8d99242e9e069b604d0e9b969e77f7dd501d";
 const MUSE_MERGE_COMMIT = "6adad08767583f52eb4d2122111af0bf638ed5e6";
+const REVISION_FIX_COMMIT = "90984ddbed27a09409506d6d6c0eea87f54b04b5";
+const MODEL_REVISION = "d35cb79050f419c457611b1cee5c5d15b176f285";
 const PIPELINE_ID = "019d130e-464e-4ff7-b84b-492992c0c06b";
-const PIPELINE_URL = "https://buildkite.com/vllm/release-v2/builds/5174";
 
 const EXPECTED_PROVENANCE = {
   schemaVersion: 1,
@@ -30,42 +31,59 @@ const EXPECTED_PROVENANCE = {
     manifestUrl: `https://registry-1.docker.io/v2/vllm/vllm-openai/manifests/${MANIFEST_DIGEST}`,
     configDigest: CONFIG_DIGEST,
     configMediaType: "application/vnd.docker.container.image.v1+json",
-    configSizeBytes: 34_762,
+    configSizeBytes: 34_738,
     configUrl: `https://registry-1.docker.io/v2/vllm/vllm-openai/blobs/${CONFIG_DIGEST}`,
     layerCount: 32,
-    compressedLayerSizeBytes: 9_699_710_136,
-    createdAt: "2026-08-14T05:33:50.528328374Z",
+    compressedLayerSizeBytes: 9_706_339_423,
+    createdAt: "2026-08-19T05:34:00.562836767Z",
     platform: {
       os: "linux",
       architecture: "arm64",
     },
   },
   build: {
-    sourceRepository: "https://github.com/vllm-project/vllm",
     sourceRevision: SOURCE_REVISION,
-    sourceRevisionUrl: `https://github.com/vllm-project/vllm/commit/${SOURCE_REVISION}`,
     imageTag: `vllm/vllm-openai:nightly-${SOURCE_REVISION}`,
     pipelineId: PIPELINE_ID,
-    pipelineUrl: PIPELINE_URL,
   },
   upstreamSupport: {
     museMergeCommit: MUSE_MERGE_COMMIT,
-    museMergeCommitUrl: `https://github.com/vllm-project/vllm/commit/${MUSE_MERGE_COMMIT}`,
-    comparisonUrl: `https://github.com/vllm-project/vllm/compare/${MUSE_MERGE_COMMIT}...${SOURCE_REVISION}`,
-    relationship: "direct-descendant",
-    aheadBy: 1,
+    relationship: "descendant",
+    aheadBy: 173,
+    revisionFixCommit: REVISION_FIX_COMMIT,
+    revisionFixRelationship: "descendant",
+    revisionFixAheadBy: 33,
   },
   reportedLabels: {
     "ai.vllm.build.commit": SOURCE_REVISION,
     "ai.vllm.build.pipeline": PIPELINE_ID,
-    "ai.vllm.build.url": PIPELINE_URL,
     "ai.vllm.image.tag": `vllm/vllm-openai:nightly-${SOURCE_REVISION}`,
     "org.opencontainers.image.revision": SOURCE_REVISION,
-    "org.opencontainers.image.source": "https://github.com/vllm-project/vllm",
+  },
+  runtimeDependencies: {
+    vllmVersion: "0.26.1rc1.dev942+g5a4c8d992",
+    huggingfaceHubVersion: "1.28.0",
+    sentencepieceVersion: "0.2.2",
+    tiktokenVersion: "0.14.0",
+  },
+  revisionSerialization: {
+    model: "Inferact/Muse-Glimmer-30B-NVFP4-W4A4",
+    requestedRevision: MODEL_REVISION,
+    resolvedRevisionBeforePickle: MODEL_REVISION,
+    resolvedRevisionAfterPickle: MODEL_REVISION,
+    preserved: true,
   },
   verification: {
-    observedAt: "2026-08-14T16:33:50Z",
-    methods: ["docker-buildx-imagetools-inspect-raw", "docker-image-inspect", "github-compare-api"],
+    observedAt: "2026-08-19T22:53:46Z",
+    methods: [
+      "docker-buildx-imagetools-inspect-raw",
+      "docker-image-inspect",
+      "github-compare-api",
+      "container-package-imports",
+      "resolved-revision-pickle-round-trip",
+      "dgx-spark-cold-cache-startup",
+      "vllm-openai-api-validation",
+    ],
     signedProvenanceAttestation: "not-available",
   },
 } as const;