diff --git a/.buildkite/test-amd.yaml b/.buildkite/test-amd.yaml
index b3d20caab..062de8f0f 100644
--- a/.buildkite/test-amd.yaml
+++ b/.buildkite/test-amd.yaml
@@ -1107,18 +1107,6 @@ steps:
   commands:
     - pytest -v -s models/quantization
 
-# This test is used only in PR development phase to test individual models and should never run on main
-- label: Custom Models Test
-  mirror_hardwares: [amdexperimental, amdproduction]
-  agent_pool: mi325_1
-  # grade: Blocking
-  optional: true
-  commands:
-    - echo 'Testing custom models...'
-    # PR authors can temporarily add commands below to test individual models
-    # e.g. pytest -v -s models/encoder_decoder/vision_language/test_mllama.py
-    # *To avoid merge conflicts, remember to REMOVE (not just comment out) them before merging the PR*
-
 - label: Transformers Nightly Models Test
   mirror_hardwares: [amdexperimental]
   agent_pool: mi325_1
@@ -1709,7 +1697,6 @@ steps:
   # in /vllm/tools/pre_commit/generate_nightly_torch_test.py
   mirror_hardwares: [amdexperimental, amdproduction, amdtentative]
   agent_pool: mi355_1
-  grade: Blocking
   soft_fail: true
   source_file_dependencies:
   - requirements/nightly_torch_test.txt
@@ -1720,7 +1707,6 @@ steps:
   timeout_in_minutes: 15
   mirror_hardwares: [amdexperimental, amdproduction, amdtentative]
   agent_pool: mi355_1
-  grade: Blocking
   source_file_dependencies:
   - vllm/
   - tests/multimodal
@@ -1733,7 +1719,6 @@ steps:
   timeout_in_minutes: 30
   mirror_hardwares: [amdexperimental, amdproduction, amdtentative]
   agent_pool: mi355_1
-  grade: Blocking
   source_file_dependencies:
   - vllm/
   - tests/test_inputs.py
@@ -1763,7 +1748,6 @@ steps:
   timeout_in_minutes: 20
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_1
-  # grade: Blocking
   source_file_dependencies:
   - tests/standalone_tests/python_only_compile.sh
   - setup.py
@@ -1774,7 +1758,6 @@ steps:
   timeout_in_minutes: 30
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_1
-  # grade: Blocking
   fast_check: true
   torch_nightly: true
   source_file_dependencies:
@@ -1791,7 +1774,6 @@ steps:
 - label: Entrypoints Unit Tests # 5min
   mirror_hardwares: [amdexperimental, amdproduction, amdtentative]
   agent_pool: mi355_1
-  grade: Blocking
   timeout_in_minutes: 10
   working_dir: "/vllm-workspace/tests"
   fast_check: true
@@ -1806,7 +1788,6 @@ steps:
   timeout_in_minutes: 40
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_1
-  # grade: Blocking
   working_dir: "/vllm-workspace/tests"
   fast_check: true
   torch_nightly: true
@@ -1824,7 +1805,6 @@ steps:
   timeout_in_minutes: 130
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_1
-  # grade: Blocking
   working_dir: "/vllm-workspace/tests"
   fast_check: true
   torch_nightly: true
@@ -1841,7 +1821,6 @@ steps:
   timeout_in_minutes: 50
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_1
-  # grade: Blocking
   working_dir: "/vllm-workspace/tests"
   fast_check: true
   torch_nightly: true
@@ -1860,7 +1839,6 @@ steps:
   timeout_in_minutes: 50
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_1
-  # grade: Blocking
   working_dir: "/vllm-workspace/tests"
   fast_check: true
   torch_nightly: true
@@ -1875,7 +1853,6 @@ steps:
   timeout_in_minutes: 50
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_1
-  # grade: Blocking
   working_dir: "/vllm-workspace/tests"
   fast_check: true
   torch_nightly: true
@@ -1890,7 +1867,6 @@ steps:
   timeout_in_minutes: 50
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_4
-  # grade: Blocking
   working_dir: "/vllm-workspace/tests"
   num_gpus: 4
   source_file_dependencies:
@@ -1952,7 +1928,6 @@ steps:
   timeout_in_minutes: 10
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_8
-  # grade: Blocking
   gpu: h100
   num_gpus: 8
   working_dir: "/vllm-workspace/tests"
@@ -1973,7 +1948,6 @@ steps:
 - label: EPLB Algorithm Test # 5min
   mirror_hardwares: [amdexperimental, amdproduction, amdtentative]
   agent_pool: mi355_1
-  grade: Blocking
   timeout_in_minutes: 15
   working_dir: "/vllm-workspace/tests"
   source_file_dependencies:
@@ -1985,7 +1959,6 @@ steps:
 - label: EPLB Execution Test # 10min
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_4
-  # grade: Blocking
   timeout_in_minutes: 20
   working_dir: "/vllm-workspace/tests"
   num_gpus: 4
@@ -2000,7 +1973,6 @@ steps:
   timeout_in_minutes: 20
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_2
-  # grade: Blocking
   num_gpus: 2
   source_file_dependencies:
   - vllm/
@@ -2020,7 +1992,6 @@ steps:
   timeout_in_minutes: 20
   mirror_hardwares: [amdexperimental, amdproduction, amdtentative]
   agent_pool: mi355_1
-  grade: Blocking
   source_file_dependencies:
   - vllm/
   - tests/test_regression
@@ -2033,7 +2004,6 @@ steps:
   timeout_in_minutes: 15
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_1
-  # grade: Blocking
   source_file_dependencies:
   - vllm/
   - tests/engine
@@ -2050,7 +2020,6 @@ steps:
   # The test uses 4 GPUs, but we schedule it on 8-GPU machines for stability.
   # See discussion here: https://github.com/vllm-project/vllm/pull/31040
   agent_pool: mi355_8
-  # grade: Blocking
   source_file_dependencies:
     - vllm/
     - tests/v1
@@ -2064,7 +2033,6 @@ steps:
   timeout_in_minutes: 50
   mirror_hardwares: [amdexperimental, amdproduction, amdtentative]
   agent_pool: mi355_1
-  grade: Blocking
   source_file_dependencies:
     - vllm/
     - tests/v1
@@ -2075,7 +2043,6 @@ steps:
   timeout_in_minutes: 60
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_1
-  # grade: Blocking
   source_file_dependencies:
     - vllm/
     - tests/v1
@@ -2103,7 +2070,6 @@ steps:
 - label: V1 Test attention (H100) # 10min
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_1
-  # grade: Blocking
   timeout_in_minutes: 30
   gpu: h100
   source_file_dependencies:
@@ -2143,7 +2109,6 @@ steps:
 - label: V1 Test others (CPU) # 5 mins
   mirror_hardwares: [amdexperimental, amdproduction, amdtentative]
   agent_pool: mi355_1
-  grade: Blocking
   source_file_dependencies:
     - vllm/
     - tests/v1
@@ -2161,7 +2126,6 @@ steps:
   timeout_in_minutes: 45
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_1
-  # grade: Blocking
   working_dir: "/vllm-workspace/examples"
   source_file_dependencies:
   - vllm/entrypoints
@@ -2196,7 +2160,6 @@ steps:
   timeout_in_minutes: 15
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_1
-  # grade: Blocking
   source_file_dependencies:
   - vllm/
   - tests/cuda
@@ -2208,7 +2171,6 @@ steps:
   timeout_in_minutes: 75
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_1
-  # grade: Blocking
   source_file_dependencies:
   - vllm/model_executor/layers
   - vllm/sampling_metadata.py
@@ -2221,7 +2183,6 @@ steps:
   timeout_in_minutes: 30
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_1
-  # grade: Blocking
   source_file_dependencies:
   - vllm/lora
   - tests/lora
@@ -2242,7 +2203,6 @@ steps:
   timeout_in_minutes: 30
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_1
-  # grade: Blocking
   torch_nightly: true
   source_file_dependencies:
     - vllm/
@@ -2259,7 +2219,6 @@ steps:
   timeout_in_minutes: 30
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_1
-  # grade: Blocking
   torch_nightly: true
   source_file_dependencies:
   - vllm/
@@ -2305,7 +2264,6 @@ steps:
   timeout_in_minutes: 75
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_1
-  # grade: Blocking
   source_file_dependencies:
   - csrc/
   - tests/kernels/core
@@ -2317,7 +2275,6 @@ steps:
   timeout_in_minutes: 35
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_1
-  # grade: Blocking
   source_file_dependencies:
   - csrc/attention/
   - vllm/v1/attention
@@ -2332,7 +2289,6 @@ steps:
   timeout_in_minutes: 90
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_1
-  # grade: Blocking
   source_file_dependencies:
   - csrc/quantization/
   - vllm/model_executor/layers/quantization
@@ -2345,7 +2301,6 @@ steps:
   timeout_in_minutes: 60
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_1
-  # grade: Blocking
   source_file_dependencies:
   - csrc/quantization/cutlass_w8a8/moe/
   - csrc/moe/
@@ -2362,7 +2317,6 @@ steps:
   timeout_in_minutes: 45
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_1
-  # grade: Blocking
   source_file_dependencies:
   - csrc/mamba/
   - tests/kernels/mamba
@@ -2406,7 +2360,6 @@ steps:
   torch_nightly: true
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_1
-  # grade: Blocking
   source_file_dependencies:
   - vllm/engine/arg_utils.py
   - vllm/config/model.py
@@ -2423,7 +2376,6 @@ steps:
   timeout_in_minutes: 20
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_1
-  # grade: Blocking
   working_dir: "/vllm-workspace/.buildkite"
   source_file_dependencies:
   - benchmarks/
@@ -2434,7 +2386,6 @@ steps:
   timeout_in_minutes: 20
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_1
-  # grade: Blocking
   source_file_dependencies:
   - vllm/
   - tests/benchmarks/
@@ -2445,7 +2396,6 @@ steps:
   timeout_in_minutes: 90
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_1
-  # grade: Blocking
   source_file_dependencies:
   - csrc/
   - vllm/model_executor/layers/quantization
@@ -2466,7 +2416,6 @@ steps:
   timeout_in_minutes: 75
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_1
-  # grade: Blocking
   source_file_dependencies:
   - csrc/
   - vllm/model_executor/layers/quantization
@@ -2478,7 +2427,6 @@ steps:
   timeout_in_minutes: 15
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_1
-  # grade: Blocking
   source_file_dependencies:
   - csrc/
   - vllm/entrypoints/openai/
@@ -2495,7 +2443,6 @@ steps:
   timeout_in_minutes: 45
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_1
-  # grade: Blocking
   torch_nightly: true
   source_file_dependencies:
   - vllm/
@@ -2508,7 +2455,6 @@ steps:
   timeout_in_minutes: 45
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_1
-  # grade: Blocking
   torch_nightly: true
   source_file_dependencies:
   - vllm/model_executor/models/
@@ -2528,7 +2474,6 @@ steps:
   timeout_in_minutes: 45
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_1
-  # grade: Blocking
   torch_nightly: true
   source_file_dependencies:
   - vllm/
@@ -2541,7 +2486,6 @@ steps:
 - label: Basic Models Test (Other CPU) # 5min
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_1
-  # grade: Blocking
   timeout_in_minutes: 10
   torch_nightly: true
   source_file_dependencies:
@@ -2556,7 +2500,6 @@ steps:
   timeout_in_minutes: 25
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_1
-  # grade: Blocking
   torch_nightly: true
   source_file_dependencies:
   - vllm/
@@ -2570,7 +2513,6 @@ steps:
   timeout_in_minutes: 45
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_1
-  # grade: Blocking
   torch_nightly: true
   source_file_dependencies:
   - vllm/model_executor/models/
@@ -2591,7 +2533,6 @@ steps:
   timeout_in_minutes: 75
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_1
-  # grade: Blocking
   torch_nightly: true
   source_file_dependencies:
   - vllm/
@@ -2612,7 +2553,6 @@ steps:
   timeout_in_minutes: 110
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_1
-  # grade: Blocking
   optional: true
   source_file_dependencies:
   - vllm/
@@ -2628,7 +2568,6 @@ steps:
   timeout_in_minutes: 110
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_1
-  # grade: Blocking
   optional: true
   source_file_dependencies:
   - vllm/
@@ -2640,7 +2579,6 @@ steps:
   timeout_in_minutes: 50
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_1
-  # grade: Blocking
   optional: true
   source_file_dependencies:
   - vllm/
@@ -2676,7 +2614,6 @@ steps:
   timeout_in_minutes: 60
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_1
-  # grade: Blocking
   source_file_dependencies:
   - vllm/
   - tests/models/multimodal
@@ -2688,7 +2625,6 @@ steps:
   timeout_in_minutes: 100
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_1
-  # grade: Blocking
   torch_nightly: true
   source_file_dependencies:
   - vllm/
@@ -2706,7 +2642,6 @@ steps:
   timeout_in_minutes: 10
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_1
-  # grade: Blocking
   working_dir: "/vllm-workspace/.buildkite/lm-eval-harness"
   source_file_dependencies:
   - vllm/multimodal/
@@ -2721,7 +2656,6 @@ steps:
   timeout_in_minutes: 120
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_1
-  # grade: Blocking
   optional: true
   source_file_dependencies:
   - vllm/
@@ -2736,7 +2670,6 @@ steps:
   timeout_in_minutes: 120
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_1
-  # grade: Blocking
   optional: true
   source_file_dependencies:
   - vllm/
@@ -2751,7 +2684,6 @@ steps:
   timeout_in_minutes: 150
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_1
-  # grade: Blocking
   optional: true
   source_file_dependencies:
   - vllm/
@@ -2766,29 +2698,15 @@ steps:
   timeout_in_minutes: 60
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_1
-  # grade: Blocking
   source_file_dependencies:
   - vllm/model_executor/layers/quantization
   - tests/models/quantization
   commands:
     - pytest -v -s models/quantization
 
-# This test is used only in PR development phase to test individual models and should never run on main
-- label: Custom Models Test
-  mirror_hardwares: [amdexperimental, amdproduction]
-  agent_pool: mi355_1
-  # grade: Blocking
-  optional: true
-  commands:
-    - echo 'Testing custom models...'
-    # PR authors can temporarily add commands below to test individual models
-    # e.g. pytest -v -s models/encoder_decoder/vision_language/test_mllama.py
-    # *To avoid merge conflicts, remember to REMOVE (not just comment out) them before merging the PR*
-
 - label: Transformers Nightly Models Test
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_1
-  # grade: Blocking
   working_dir: "/vllm-workspace/"
   optional: true
   commands:
@@ -2927,7 +2845,6 @@ steps:
   timeout_in_minutes: 20
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_2
-  # grade: Blocking
   working_dir: "/vllm-workspace/tests"
   num_gpus: 2
   source_file_dependencies:
@@ -2943,7 +2860,6 @@ steps:
   timeout_in_minutes: 30
   mirror_hardwares: [amdexperimental, amdmultinode]
   agent_pool: mi355_4
-  # grade: Blocking
   working_dir: "/vllm-workspace/tests"
   num_gpus: 2
   num_nodes: 2
@@ -2970,7 +2886,6 @@ steps:
   timeout_in_minutes: 90
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_2
-  # grade: Blocking
   working_dir: "/vllm-workspace/tests"
   num_gpus: 2
   source_file_dependencies:
@@ -3010,7 +2925,6 @@ steps:
   timeout_in_minutes: 50
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_2
-  # grade: Blocking
   working_dir: "/vllm-workspace/tests"
   num_gpus: 2
   source_file_dependencies:
@@ -3032,7 +2946,6 @@ steps:
   timeout_in_minutes: 60
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_2
-  # grade: Blocking
   working_dir: "/vllm-workspace/tests"
   num_gpus: 2
   source_file_dependencies:
@@ -3066,7 +2979,6 @@ steps:
   timeout_in_minutes: 60
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_4
-  # grade: Blocking
   working_dir: "/vllm-workspace/tests"
   num_gpus: 4
   source_file_dependencies:
@@ -3083,7 +2995,6 @@ steps:
   timeout_in_minutes: 30
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_4
-  # grade: Blocking
   num_gpus: 4
   source_file_dependencies:
   - vllm/lora
@@ -3108,7 +3019,6 @@ steps:
   timeout_in_minutes: 45
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_2
-  # grade: Blocking
   working_dir: "/vllm-workspace/tests"
   num_gpus: 2
   optional: true
@@ -3121,7 +3031,6 @@ steps:
 - label: Weight Loading Multiple GPU Test - Large Models # optional
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_2
-  # grade: Blocking
   working_dir: "/vllm-workspace/tests"
   num_gpus: 2
   optional: true
@@ -3134,7 +3043,6 @@ steps:
 - label: NixlConnector PD accuracy tests (Distributed) # 30min
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_4
-  # grade: Blocking
   timeout_in_minutes: 30
   working_dir: "/vllm-workspace/tests"
   num_gpus: 4
@@ -3148,7 +3056,6 @@ steps:
 - label: DP EP NixlConnector PD accuracy tests (Distributed) # 15min
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_4
-  # grade: Blocking
   timeout_in_minutes: 15
   working_dir: "/vllm-workspace/tests"
   num_gpus: 4
@@ -3165,7 +3072,6 @@ steps:
 - label: Distributed Tests (A100) # optional
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_4
-  # grade: Blocking
   gpu: a100
   optional: true
   num_gpus: 4
@@ -3188,7 +3094,6 @@ steps:
   optional: true
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_4
-  # grade: Blocking
   num_gpus: 4
   working_dir: "/vllm-workspace/.buildkite/lm-eval-harness"
   source_file_dependencies:
@@ -3204,7 +3109,6 @@ steps:
   optional: true
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_4
-  # grade: Blocking
   num_gpus: 4
   working_dir: "/vllm-workspace/.buildkite/lm-eval-harness"
   source_file_dependencies:
@@ -3219,7 +3123,6 @@ steps:
 - label: Distributed Tests (H200) # optional
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_2
-  # grade: Blocking
   gpu: h200
   optional: true
   working_dir: "/vllm-workspace/"
@@ -3254,7 +3157,6 @@ steps:
   timeout_in_minutes: 20
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_1
-  # grade: Blocking
   source_file_dependencies:
   - csrc/
   - vllm/model_executor/layers/quantization
@@ -3264,7 +3166,6 @@ steps:
 - label: LM Eval Large Models (4 Card)
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_4
-  # grade: Blocking
   gpu: a100
   optional: true
   num_gpus: 4
@@ -3304,7 +3205,6 @@ steps:
 - label: Prime-RL Integration Test # 15min
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_2
-  # grade: Blocking
   timeout_in_minutes: 30
   optional: true
   num_gpus: 2
@@ -3319,7 +3219,6 @@ steps:
 - label: DeepSeek V2-Lite Accuracy
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_4
-  # grade: Blocking
   timeout_in_minutes: 60
   gpu: h100
   optional: true
@@ -3331,7 +3230,6 @@ steps:
 - label: Qwen3-30B-A3B-FP8-block Accuracy (H100)
   mirror_hardwares: [amdexperimental, amdproduction]
   agent_pool: mi355_4
-  # grade: Blocking
   timeout_in_minutes: 60
   gpu: h100
   optional: true
@@ -3354,7 +3252,6 @@ steps:
   timeout_in_minutes: 60
   mirror_hardwares: [amdexperimental]
   agent_pool: mi355_4
-  # grade: Blocking
   optional: true
   num_gpus: 4
   working_dir: "/vllm-workspace"