diff --git a/cmake/external_projects/vllm_flash_attn.cmake b/cmake/external_projects/vllm_flash_attn.cmake
index b51934a3ab29..dbdfd5e81443 100644
--- a/cmake/external_projects/vllm_flash_attn.cmake
+++ b/cmake/external_projects/vllm_flash_attn.cmake
@@ -38,7 +38,7 @@ else()
   FetchContent_Declare(
           vllm-flash-attn
           GIT_REPOSITORY https://github.com/vllm-project/flash-attention.git
-          GIT_TAG 188be16520ceefdc625fdf71365585d2ee348fe2
+          GIT_TAG 2adfc8c2177c5b0e8ddeedfd5a8990d80eb496ff
           GIT_PROGRESS TRUE
           # Don't share the vllm-flash-attn build between build types
           BINARY_DIR ${CMAKE_BINARY_DIR}/vllm-flash-attn
diff --git a/vllm/v1/attention/backends/flash_attn.py b/vllm/v1/attention/backends/flash_attn.py
index 9275725314e4..232b0b0daff6 100755
--- a/vllm/v1/attention/backends/flash_attn.py
+++ b/vllm/v1/attention/backends/flash_attn.py
@@ -308,10 +308,15 @@ def __init__(
             self.compilation_config.cudagraph_mode.has_full_cudagraphs()
         )
         self.max_cudagraph_size = self.compilation_config.max_cudagraph_capture_size
+        max_num_seqs = vllm_config.scheduler_config.max_num_seqs
 
         if self.use_full_cuda_graph and self.aot_schedule:
+            # Times 4 due to:
+            #  https://github.com/vllm-project/flash-attention/blob/3223650ccabe622a0fcae65eec706a50186a89f7/hopper/flash_api.cpp#L650-L653
+            # For some tests max_cudagraph_size > max_num_seqs,
+            #   so we need to use the larger one.
             self.scheduler_metadata = torch.zeros(
-                vllm_config.scheduler_config.max_num_seqs + 1,
+                max(self.max_cudagraph_size or 0, max_num_seqs) * 4 + 1,
                 dtype=torch.int32,
                 device=self.device,
             )
diff --git a/vllm/v1/attention/backends/mla/flashattn_mla.py b/vllm/v1/attention/backends/mla/flashattn_mla.py
index e160d3255688..f0ba259362ff 100644
--- a/vllm/v1/attention/backends/mla/flashattn_mla.py
+++ b/vllm/v1/attention/backends/mla/flashattn_mla.py
@@ -127,10 +127,15 @@ def __init__(
             self.compilation_config.cudagraph_mode.has_full_cudagraphs()
         )
         self.max_cudagraph_size = self.compilation_config.max_cudagraph_capture_size
+        max_num_seqs = vllm_config.scheduler_config.max_num_seqs
 
         if self.use_full_cuda_graph and self.fa_aot_schedule:
+            # Times 4 due to:
+            #  https://github.com/vllm-project/flash-attention/blob/3223650ccabe622a0fcae65eec706a50186a89f7/hopper/flash_api.cpp#L650-L653
+            # For some tests max_cudagraph_size > max_num_seqs,
+            #   so we need to use the larger one.
             self.scheduler_metadata = torch.zeros(
-                vllm_config.scheduler_config.max_num_seqs + 1,
+                max(self.max_cudagraph_size or 0, max_num_seqs) * 4 + 1,
                 dtype=torch.int32,
                 device=self.device,
             )