diff --git a/vllm/model_executor/layers/quantization/utils/flashinfer_fp4_moe.py b/vllm/model_executor/layers/quantization/utils/flashinfer_fp4_moe.py
index ddb74a27dc122..5ce0188b60aed 100644
--- a/vllm/model_executor/layers/quantization/utils/flashinfer_fp4_moe.py
+++ b/vllm/model_executor/layers/quantization/utils/flashinfer_fp4_moe.py
@@ -32,7 +32,7 @@ def is_flashinfer_fp4_cutlass_moe_available() -> bool:
         envs.VLLM_USE_FLASHINFER_MOE_FP4
         and has_flashinfer_cutlass_fused_moe()
         and current_platform.is_cuda()
-        and current_platform.is_device_capability(100)
+        and current_platform.has_device_capability(100)
     )