diff --git a/vllm/config/parallel.py b/vllm/config/parallel.py
index f7b8b1d0a565..9ea883d4a03c 100644
--- a/vllm/config/parallel.py
+++ b/vllm/config/parallel.py
@@ -6,7 +6,7 @@
 from typing import TYPE_CHECKING, Any, Literal, Optional, Union
 
 import torch
-from pydantic import TypeAdapter, model_validator
+from pydantic import model_validator
 from pydantic.dataclasses import dataclass
 from torch.distributed import ProcessGroup, ReduceOp
 from typing_extensions import Self
@@ -56,13 +56,6 @@ class EPLBConfig:
     This is turned off by default since it will cause communication overhead.
     """
 
-    @classmethod
-    def from_cli(cls, cli_value: str) -> "EPLBConfig":
-        """Parse the CLI value for the compilation config.
-        -O1, -O2, -O3, etc. is handled in FlexibleArgumentParser.
-        """
-        return TypeAdapter(EPLBConfig).validate_json(cli_value)
-
 
 @config
 @dataclass
diff --git a/vllm/engine/arg_utils.py b/vllm/engine/arg_utils.py
index 965264ee3097..3ab1115f1446 100644
--- a/vllm/engine/arg_utils.py
+++ b/vllm/engine/arg_utils.py
@@ -455,8 +455,7 @@ def __post_init__(self):
             self.compilation_config = CompilationConfig(
                 **self.compilation_config)
         if isinstance(self.eplb_config, dict):
-            self.eplb_config = EPLBConfig.from_cli(json.dumps(
-                self.eplb_config))
+            self.eplb_config = EPLBConfig(**self.eplb_config)
         # Setup plugins
         from vllm.plugins import load_general_plugins
         load_general_plugins()