diff --git a/vllm/transformers_utils/config.py b/vllm/transformers_utils/config.py
index fd76550d664..f900eedf8d3 100644
--- a/vllm/transformers_utils/config.py
+++ b/vllm/transformers_utils/config.py
@@ -72,6 +72,7 @@ class LazyConfigDict(dict):
 _CONFIG_REGISTRY: dict[str, type[PretrainedConfig]] = LazyConfigDict(
     afmoe="AfmoeConfig",
     arctic="ArcticConfig",
+    axk1="AXK1Config",
     bagel="BagelConfig",
     umm="CheersConfig",
     chatglm="ChatGLMConfig",
diff --git a/vllm/transformers_utils/configs/AXK1.py b/vllm/transformers_utils/configs/AXK1.py
index 5c19a37324b..09a3a0f46ae 100644
--- a/vllm/transformers_utils/configs/AXK1.py
+++ b/vllm/transformers_utils/configs/AXK1.py
@@ -114,7 +114,7 @@ class AXK1Config(PretrainedConfig):
             The dropout ratio for the attention probabilities.
     """
 
-    model_type = "AXK1"
+    model_type = "axk1"
     keys_to_ignore_at_inference = ["past_key_values"]
 
     def __init__(
diff --git a/vllm/transformers_utils/model_arch_config_convertor.py b/vllm/transformers_utils/model_arch_config_convertor.py
index bd146dff7dc..70bb4caa535 100644
--- a/vllm/transformers_utils/model_arch_config_convertor.py
+++ b/vllm/transformers_utils/model_arch_config_convertor.py
@@ -261,7 +261,7 @@ class ModelArchConfigConvertorBase:
         if not hasattr(self.hf_text_config, "model_type"):
             return False
         elif self.hf_text_config.model_type in (
-            "AXK1",
+            "axk1",
             "deepseek_v2",
             "deepseek_v3",
             "deepseek_v32",
@@ -290,7 +290,7 @@ class ModelArchConfigConvertorBase:
             return (
                 self.hf_text_config.model.model_type
                 in (
-                    "AXK1",
+                    "axk1",
                     "deepseek_v2",
                     "deepseek_v3",
                     "deepseek_v32",