diff --git a/computility-run.yaml b/computility-run.yaml
index 226cd7f1..568b4ca1 100644
--- a/computility-run.yaml
+++ b/computility-run.yaml
@@ -49,3 +49,5 @@ env:
       value: '1'
     - name: PYTORCH_CUDA_ALLOC_CONF
       value: max_split_size_mb:512
+    - name: BI100_MAX_GPU_BLOCKS
+      value: '5000'
diff --git a/qwen3_6_scripts/patch_block_major_worker_capacity.py b/qwen3_6_scripts/patch_block_major_worker_capacity.py
index 599e194e..7ef25aa0 100644
--- a/qwen3_6_scripts/patch_block_major_worker_capacity.py
+++ b/qwen3_6_scripts/patch_block_major_worker_capacity.py
@@ -20,6 +20,14 @@ CAPACITY_ANCHOR = """\
 CAPACITY_REPLACEMENT = """\
         num_gpu_blocks = reserve_block_major_gpu_blocks(
             num_gpu_blocks, cache_block_size)
+        # BI100: cap GPU blocks — profiling with zero-tensor attention
+        # underestimates memory, causing runtime OOM if uncapped.
+        _bi100_max = int(os.environ.get("BI100_MAX_GPU_BLOCKS", "0"))
+        if _bi100_max > 0 and num_gpu_blocks > _bi100_max:
+            logger.warning(
+                "[BI100] capping num_gpu_blocks: %d -> %d (BI100_MAX_GPU_BLOCKS)",
+                num_gpu_blocks, _bi100_max)
+            num_gpu_blocks = _bi100_max
         num_gpu_blocks = max(num_gpu_blocks, 0)
         num_cpu_blocks = max(num_cpu_blocks, 0)
 """