diff --git a/.github/workflows/vmbench-long-range.yml b/.github/workflows/vmbench-long-range.yml
index 0c1dd67971e..625156b6652 100644
--- a/.github/workflows/vmbench-long-range.yml
+++ b/.github/workflows/vmbench-long-range.yml
@@ -56,7 +56,7 @@ on:
         default: v3.2.1@sha256:6927e0919a144aa7616fd0137d4816816d42f6b816de3af269ab065250859a62
         required: true
       runtime_image:
-        description: 'Runtime image: auto = pinned S3 test runtime on ACR; explicit tag/digest overrides'
+        description: 'Runtime image: auto = pinned Prometheus/File/S3 runtime on ACR; explicit tag/digest overrides'
         type: string
         default: auto
         required: true
@@ -116,7 +116,7 @@ env:
   GREPTIMEDB_TAG: ${{ inputs.greptimedb_tag }}
   VICTORIAMETRICS_TAG: ${{ inputs.victoriametrics_tag }}
   PROMETHEUS_TAG: ${{ inputs.prometheus_tag }}
-  RUNTIME_IMAGE: ${{ inputs.runtime_image != 'auto' && inputs.runtime_image || 'greptime-registry.cn-hangzhou.cr.aliyuncs.com/tools/o11ybench-runtime:s3-test-b0cf5204d2fc@sha256:c2bf622db6056a2029842f4dc1cc1f37b82773bd45b134a0a83ffe2a8cd62214' }}
+  RUNTIME_IMAGE: ${{ inputs.runtime_image != 'auto' && inputs.runtime_image || 'greptime-registry.cn-hangzhou.cr.aliyuncs.com/tools/o11ybench-runtime:345b6f7c-prometheus-s3@sha256:39d207f4a9c479b8614b8283e2d049fc8f430fcbd2b3362ce3035cfbf294edf9' }}
   PROVIDER: ${{ inputs.provider }}
   REQUESTED_INSTANCE_TYPE: ${{ inputs.instance_type }}
   SYSTEM_DISK_GIB: ${{ inputs.system_disk_gib }}
diff --git a/tests/perf/test_o11ybench_storage.py b/tests/perf/test_o11ybench_storage.py
index 308469fd165..d1bd76a3771 100644
--- a/tests/perf/test_o11ybench_storage.py
+++ b/tests/perf/test_o11ybench_storage.py
@@ -45,13 +45,15 @@ class StorageTests(unittest.TestCase):
         plan = dict(output_root='/tmp/o11ybench-test', run_id='test',
                     data=dict(generate=True, data_home='/tmp/o11ybench-test/data'),
                     targets=[dict(name='greptimedb', storage=dict(type='file')),
-                             dict(name='greptimedb-s3', storage=dict(type='s3', root='owned/prefix/'))],
+                             dict(name='greptimedb-s3', storage=dict(type='s3', root='owned/prefix/')),
+                             dict(name='prometheus', kind='prometheus')],
                     report=dict(root='old', index_html='old/index.html'),
                     artifacts=dict(run_manifest='old/run-manifest.json'),
                     query=dict(modes=['serial'], serial_tries=5))
         original = copy.deepcopy(plan)
         children = storage.split_metrics(plan)
         self.assertEqual(plan, original)
+        self.assertEqual(set(children), {'greptimedb', 'greptimedb-s3', 'prometheus'})
         for name, child in children.items():
             self.assertEqual(child['query'], plan['query'])
             self.assertEqual(child['data'], dict(plan['data'], generate=False))
@@ -61,6 +63,15 @@ class StorageTests(unittest.TestCase):
             self.assertTrue(child['report']['root'].startswith(child['output_root']))
         self.assertNotEqual(children['greptimedb']['run_id'], children['greptimedb-s3']['run_id'])
 
+    def test_prometheus_workflow_wiring(self):
+        text = (ROOT / '.github/workflows/vmbench-long-range.yml').read_text()
+        step = text.split('    - name: Load and benchmark Prometheus\n', 1)[1].split('    - name:', 1)[0]
+        self.assertIn("steps.generate.outcome == 'success' && inputs.run_prometheus", step)
+        self.assertIn('metrics-run --target prometheus', step)
+        self.assertIn('for target in greptimedb victoriametrics prometheus;', text)
+        cleanup = text.split('    - name: Clean up owned containers', 1)[1].split('    - name:', 1)[0]
+        self.assertIn('greptimedb-s3-baseline victoriametrics prometheus; do', cleanup)
+
     def test_missing_s3_arm_is_not_a_success(self):
         with tempfile.TemporaryDirectory() as tmp:
             root = Path(tmp)