diff --git a/vllm/v1/attention/backends/flash_attn.py b/vllm/v1/attention/backends/flash_attn.py
index bfb4a45c2b56..81623549ae85 100755
--- a/vllm/v1/attention/backends/flash_attn.py
+++ b/vllm/v1/attention/backends/flash_attn.py
@@ -130,6 +130,12 @@ class FlashAttentionBackend(AttentionBackend):
             return flash_attn_supports_fp8()
         return kv_cache_dtype in ["auto"]
 
+    @classmethod
+    def supports_sink(cls) -> bool:
+        if not is_flash_attn_varlen_func_available():
+            return False
+        return flash_attn_supports_sinks()
+
     @classmethod
     def supports_compute_capability(cls, capability: DeviceCapability) -> bool:
         return capability >= DeviceCapability(8, 0)