diff --git a/vllm/model_executor/warmup/kernel_warmup.py b/vllm/model_executor/warmup/kernel_warmup.py
index 79d1927d3210..ffa3bc8f021e 100644
--- a/vllm/model_executor/warmup/kernel_warmup.py
+++ b/vllm/model_executor/warmup/kernel_warmup.py
@@ -11,7 +11,7 @@
 import torch
 
 import vllm.envs as envs
-from vllm.config import VllmConfig
+from vllm.config import CUDAGraphMode, VllmConfig
 from vllm.logger import init_logger
 from vllm.model_executor.warmup.deep_gemm_warmup import deep_gemm_warmup
 from vllm.platforms import current_platform
@@ -30,13 +30,19 @@ def flashinfer_autotune_supported(vllm_config: VllmConfig) -> bool:
     Record known issues with vllm + flashinfer autotune here. Return True if
     and only if flashinfer autotune will run through without issues.
     """
-    return not (
-        vllm_config.parallel_config.data_parallel_size > 1
-        and (
-            envs.VLLM_USE_FLASHINFER_MOE_MXFP4_BF16
-            or envs.VLLM_USE_FLASHINFER_MOE_MXFP4_MXFP8
-        )
+    is_tp_or_dp = (vllm_config.parallel_config.data_parallel_size > 1) or (
+        vllm_config.parallel_config.tensor_parallel_size > 1
     )
+    is_fi_mxfp4_backend = (
+        envs.VLLM_USE_FLASHINFER_MOE_MXFP4_MXFP8
+        or envs.VLLM_USE_FLASHINFER_MOE_MXFP4_BF16
+        or envs.VLLM_USE_FLASHINFER_MOE_MXFP4_MXFP8_CUTLASS
+    ) or (
+        current_platform.is_cuda() and current_platform.is_device_capability(100)
+    )  # on >=sm100, default mxfp4 backend is flashinfer
+    is_eager = vllm_config.compilation_config.cudagraph_mode == CUDAGraphMode.NONE
+
+    return not (is_tp_or_dp and is_fi_mxfp4_backend and is_eager)
 
 
 def kernel_warmup(worker: "Worker"):