diff --git a/vllm/engine/arg_utils.py b/vllm/engine/arg_utils.py
index ec61fc4b9b06e..7763dbc392e4b 100644
--- a/vllm/engine/arg_utils.py
+++ b/vllm/engine/arg_utils.py
@@ -1131,6 +1131,10 @@ class EngineArgs:
         device_config = DeviceConfig(
             device=cast(Device, current_platform.device_type))
 
+        model_config = self.create_model_config()
+        self.model = model_config.model
+        self.tokenizer = model_config.tokenizer
+
         (self.model, self.tokenizer,
          self.speculative_config) = maybe_override_with_speculators(
              model=self.model,
@@ -1139,7 +1143,6 @@ class EngineArgs:
              trust_remote_code=self.trust_remote_code,
              vllm_speculative_config=self.speculative_config,
          )
-        model_config = self.create_model_config()
 
         # * If VLLM_USE_V1 is unset, we enable V1 for "supported features"
         #   and fall back to V0 for experimental or unsupported features.