diff --git a/vllm/v1/worker/gpu/attn_utils.py b/vllm/v1/worker/gpu/attn_utils.py
index 631bcd8023526..f23b9187361fa 100644
--- a/vllm/v1/worker/gpu/attn_utils.py
+++ b/vllm/v1/worker/gpu/attn_utils.py
@@ -136,4 +136,4 @@ def init_kv_cache(
     kv_cache_raw_tensors = _allocate_kv_cache(kv_cache_config, device)
     kv_caches = _reshape_kv_cache(kv_cache_config, kv_cache_raw_tensors,
                                   attn_backends)
-    bind_kv_cache(forward_context, kv_caches, runner_kv_caches)
+    bind_kv_cache(kv_caches, forward_context, runner_kv_caches)