diff --git a/.buildkite/test_areas/lm_eval.yaml b/.buildkite/test_areas/lm_eval.yaml
index 6bbe79232c0d..941b361e50de 100644
--- a/.buildkite/test_areas/lm_eval.yaml
+++ b/.buildkite/test_areas/lm_eval.yaml
@@ -15,6 +15,7 @@ steps:
   - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=configs/models-small.yaml
   mirror:
     amd:
+      automatic_shard: true
       label: ":amd: (MI300) LM Eval Small Models"
       dind: false
       device: mi300_1
@@ -33,6 +34,7 @@ steps:
 
 - label: ":amd: (MI300) LM Eval Small Models Harness"
   key: amd-lm-eval-small-models-harness
+  automatic_shard: true
   dind: false
   device: mi300_1
   timeout_in_minutes: 35
@@ -55,6 +57,7 @@ steps:
 - label: ":nvidia: (H200 MIG 35GB) LM Eval Watermarking"
   device: h200_35gb
   key: lm-eval-watermarking
+  automatic_shard: true
   # Recorded runs took 80–140 seconds
   timeout_in_minutes: 10
   source_file_dependencies:
@@ -69,6 +72,7 @@ steps:
   - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=configs/models-watermark.txt
   mirror:
     amd:
+      automatic_shard: true
       label: ":amd: (MI355 DPX) LM Eval Watermarking"
       dind: false
       device: mi355_dpx
@@ -267,6 +271,7 @@ steps:
 
 - label: ":nvidia: (H100) LM Eval Watermark Feature Combination"
   key: lm-eval-dspark-watermark-2xh100
+  automatic_shard: true
   device: h100
   num_devices: 2
   # Local TP2 runs took 84–379 seconds, including startup.
@@ -296,6 +301,7 @@ steps:
     watermarking/test_parallel_compatibility.py
   mirror:
     amd:
+      automatic_shard: true
       label: ":amd: (MI355) LM Eval Watermark Feature Combination"
       dind: false
       device: mi355_2
@@ -678,6 +684,7 @@ steps:
 
 - label: ":nvidia: (H200) [Nightly] LM Eval Nemotron 3 Super"
   key: lm-eval-nemotron-3-super-8xh200
+  automatic_shard: true
   timeout_in_minutes: 35
   device: h200
   optional: true
@@ -686,6 +693,7 @@ steps:
     - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=configs/models-h200-nemotron-3-super.txt
   mirror:
     amd:
+      automatic_shard: true
       label: ":amd: (MI355) LM Eval Nemotron 3 Super"
       dind: false
       device: mi355_4
@@ -722,6 +730,7 @@ steps:
 
 - label: ":amd: (MI355) [Nightly] LM Eval Large Models"
   key: amd-lm-eval-large-models
+  automatic_shard: true
   dind: false
   device: mi355_8
   optional: true
@@ -745,6 +754,7 @@ steps:
     - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=evals/gsm8k/configs/moe-refactor/config-h100.txt
   mirror:
     amd:
+      automatic_shard: true
       label: ":amd: (MI355) MoE Refactor Integration"
       dind: false
       device: mi355_2
@@ -774,6 +784,7 @@ steps:
     - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=evals/gsm8k/configs/moe-refactor/config-b200.txt
   mirror:
     amd:
+      automatic_shard: true
       label: ":amd: (MI355) MoE Refactor BF16 Integration"
       dind: false
       device: mi355_2
@@ -808,6 +819,7 @@ steps:
     - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=evals/gsm8k/configs/moe-refactor-dp-ep/config-b200.txt
   mirror:
     amd:
+      automatic_shard: true
       label: ":amd: (MI355) MoE Refactor BF16 DP Integration"
       dind: false
       device: mi355_2
@@ -920,6 +932,7 @@ steps:
 
 - label: ":nvidia: (H100) [Nightly] LM Eval Humming DeepEP TEMPORARY"
   key: lm-eval-humming-h100-deepep
+  automatic_shard: true
   optional: true
   timeout_in_minutes: 60
   device: h100
@@ -972,6 +985,7 @@ steps:
 
 - label: ":nvidia: (H200 MIG 18GB) LM Eval TurboQuant k8v4"
   key: lm-eval-turboquant-k8v4
+  automatic_shard: true
   timeout_in_minutes: 30
   device: h200_18gb
   source_file_dependencies:
@@ -983,6 +997,7 @@ steps:
   - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=evals/gsm8k/configs/models-turboquant-k8v4.txt
   mirror:
     amd:
+      automatic_shard: true
       label: ":amd: (MI355 DPX) LM Eval TurboQuant k8v4"
       dind: false
       device: mi355_dpx
@@ -1133,6 +1148,7 @@ steps:
 # Octave kernels are built only for gfx942 and gfx1100, so this runs on MI300.
 - label: ":amd: (MI300) LM Eval Octave KV Cache"
   key: amd-lm-eval-octave-mi300
+  automatic_shard: true
   dind: false
   device: mi300_1
   num_devices: 1
@@ -1189,6 +1205,7 @@ steps:
 
 - label: ":nvidia: (B200) [Nightly] GPQA Eval (GPT-OSS)"
   key: gpqa-eval-gpt-oss-2xb200
+  automatic_shard: true
   timeout_in_minutes: 30
   device: b200-k8s
   optional: true
@@ -1201,6 +1218,7 @@ steps:
     - pytest -s -v evals/gpt_oss/test_gpqa_correctness.py --config-list-file=configs/models-b200.txt
   mirror:
     amd:
+      automatic_shard: true
       label: ":amd: (MI355) GPQA Eval (GPT-OSS)"
       dind: false
       device: mi355_2
@@ -1279,6 +1297,7 @@ steps:
 
 - label: ":nvidia: (DGX Spark) [Nightly] GPQA Eval (GPT-OSS)"
   key: gpqa-eval-gpt-oss-spark
+  automatic_shard: true
   timeout_in_minutes: 35
   device: dgx-spark
   optional: true
@@ -1293,6 +1312,7 @@ steps:
     - pytest -s -v evals/gpt_oss/test_gpqa_correctness.py --config-list-file=configs/models-spark.txt
   mirror:
     amd:
+      automatic_shard: true
       label: ":amd: (MI355 DPX) GPQA Eval (GPT-OSS TP1)"
       # The Spark config is a backend-neutral GPT-OSS TP1 baseline.
       dind: false
@@ -1316,6 +1336,7 @@ steps:
 
 - label: ":nvidia: (H200 MIG 35GB) LM Eval KV-Offload"
   key: kv-offload-small
+  automatic_shard: true
   timeout_in_minutes: 30
   device: h200_35gb
   source_file_dependencies:
@@ -1328,6 +1349,7 @@ steps:
   - pytest -s -v evals/gsm8k/test_gsm8k_offloading.py -k "nemotron-h-8b or gemma-4-e4b-it"
   mirror:
     amd:
+      automatic_shard: true
       label: ":amd: (MI355 DPX) LM Eval KV-Offload"
       dind: false
       device: mi355_dpx
@@ -1345,6 +1367,7 @@ steps:
 
 - label: ":nvidia: (H100) LM Eval KV-Offload Medium"
   key: kv-offload-medium
+  automatic_shard: true
   timeout_in_minutes: 45
   device: h100
   num_devices: 2
@@ -1358,6 +1381,7 @@ steps:
   - pytest -s -v evals/gsm8k/test_gsm8k_offloading.py -k "qwen3.5-35b or deepseek-v2-lite"
   mirror:
     amd:
+      automatic_shard: true
       label: ":amd: (MI355) LM Eval KV-Offload Medium"
       dind: false
       device: mi355_2
@@ -1375,6 +1399,7 @@ steps:
 
 - label: ":nvidia: (H100) [Nightly] LM Eval KV-Offload Large"
   key: kv-offload-large
+  automatic_shard: true
   # Nightly only: 4-GPU H100 pods wait behind smaller jobs on per-PR runs.
   optional: true
   timeout_in_minutes: 40
@@ -1390,6 +1415,7 @@ steps:
   - pytest -s -v evals/gsm8k/test_gsm8k_offloading.py -k "deepseek-v4-flash"
   mirror:
     amd:
+      automatic_shard: true
       label: ":amd: (MI355) LM Eval KV-Offload Large"
       dind: false
       device: mi355_4
@@ -1409,6 +1435,7 @@ steps:
 
 - label: ":nvidia: (H200 MIG 35GB) MRCR Eval Small Models"
   key: mrcr-eval-small-models
+  automatic_shard: true
   device: h200_35gb
   timeout_in_minutes: 25
   source_file_dependencies:
@@ -1417,6 +1444,7 @@ steps:
     - pytest -s -v evals/mrcr/test_mrcr_correctness.py --config-list-file=evals/mrcr/configs/models-small.txt
   mirror:
     amd:
+      automatic_shard: true
       label: ":amd: (MI355 DPX) MRCR Eval Small Models"
       dind: false
       device: mi355_dpx
@@ -1434,6 +1462,7 @@ steps:
 
 - label: ":amd: (MI355) [Nightly] LM Eval Large Models Harness"
   key: amd-lm-eval-large-models-harness
+  automatic_shard: true
   dind: false
   device: mi355_8
   optional: true
@@ -1459,6 +1488,7 @@ steps:
 
 - label: ":amd: (MI355) [Nightly] LM Eval Large Models ROCm Harness"
   key: amd-lm-eval-large-models-rocm-harness
+  automatic_shard: true
   dind: false
   device: mi355_8
   optional: true