diff --git a/.github/workflows/pr-test.yml b/.github/workflows/pr-test.yml
index 2287eea8f..db61c614e 100644
--- a/.github/workflows/pr-test.yml
+++ b/.github/workflows/pr-test.yml
@@ -820,7 +820,10 @@ jobs:
           ((needs.check-changes.outputs.main_package == 'true') || (needs.check-changes.outputs.sgl_kernel == 'true'))
         )
       )
-    runs-on: 1-gpu-h100
+    # `1-gpu-h100-h200` is a shared label advertised by both the 1-GPU H100 and
+    # 1-GPU H200 runner pools, so this job can land on either. Multimodal_gen
+    # jobs in pr-test-multimodal-gen.yml continue to pin to `1-gpu-h100`.
+    runs-on: 1-gpu-h100-h200
     timeout-minutes: 240
     strategy:
       fail-fast: false
diff --git a/scripts/ci/utils/slash_command_handler.py b/scripts/ci/utils/slash_command_handler.py
index 4f2f845a2..2f53608ec 100644
--- a/scripts/ci/utils/slash_command_handler.py
+++ b/scripts/ci/utils/slash_command_handler.py
@@ -479,7 +479,7 @@ CUDA_SUITE_TO_RUNNER = {
     "stage-a-test-1-gpu-small": "1-gpu-5090",
     "stage-a-test-cpu": "ubuntu-latest",
     "stage-b-test-1-gpu-small": "1-gpu-5090",
-    "stage-b-test-1-gpu-large": "1-gpu-h100",
+    "stage-b-test-1-gpu-large": "1-gpu-h100-h200",
     "stage-b-test-2-gpu-large": "2-gpu-h100",
     "stage-b-test-4-gpu-b200": "4-gpu-b200",
     "stage-c-test-4-gpu-h100": "4-gpu-h100",