diff --git a/vllm/v1/attention/backends/linear_attn.py b/vllm/v1/attention/backends/linear_attn.py
index 1900c50849eca..004baa2d09cde 100644
--- a/vllm/v1/attention/backends/linear_attn.py
+++ b/vllm/v1/attention/backends/linear_attn.py
@@ -7,6 +7,7 @@ import torch
 from vllm.attention.backends.abstract import AttentionBackend
 from vllm.config import VllmConfig
 from vllm.v1.attention.backends.utils import (
+    AttentionCGSupport,
     AttentionMetadataBuilder,
     CommonAttentionMetadata,
     split_decodes_and_prefills,
@@ -35,6 +36,8 @@ class LinearAttentionMetadata:
 class LinearAttentionMetadataBuilder(AttentionMetadataBuilder[LinearAttentionMetadata]):
     reorder_batch_threshold: int = 1
 
+    _cudagraph_support = AttentionCGSupport.UNIFORM_SINGLE_TOKEN_DECODE
+
     def __init__(
         self,
         kv_cache_spec: AttentionSpec,