diff --git a/vllm/config.py b/vllm/config.py
index 90d574bb3b2aa..9b92d9706c9dd 100644
--- a/vllm/config.py
+++ b/vllm/config.py
@@ -164,9 +164,6 @@ def get_num_kv_heads(self, parallel_config: "ParallelConfig") -> int:
         total_num_attention_heads = self.hf_config.num_attention_heads
         return total_num_attention_heads // parallel_config.tensor_parallel_size
 
-    def get_max_model_len(self) -> int:
-        return self.max_model_len
-
     def get_num_layers(self, parallel_config: "ParallelConfig") -> int:
         total_num_hidden_layers = self.hf_config.num_hidden_layers
         return total_num_hidden_layers // parallel_config.pipeline_parallel_size
@@ -378,10 +375,17 @@ def _get_and_verify_max_len(
         if max_len_key is not None:
             derived_max_model_len = min(derived_max_model_len, max_len_key)
     if derived_max_model_len == float("inf"):
-        raise ValueError(
-            "The model's config.json must contain one of the following keys "
-            "to determine the original maximum length of the model: "
-            f"{possible_keys}")
+        if max_model_len is not None:
+            # If max_model_len is specified, we use it.
+            return max_model_len
+
+        default_max_len = 2048
+        logger.warning(
+            "The model's config.json does not contain any of the following "
+            "keys to determine the original maximum length of the model: "
+            f"{possible_keys}. Assuming the model's maximum length is "
+            f"{default_max_len}.")
+        derived_max_model_len = default_max_len
 
     rope_scaling = getattr(hf_config, "rope_scaling", None)
     if rope_scaling is not None:
diff --git a/vllm/engine/arg_utils.py b/vllm/engine/arg_utils.py
index 8951a98e4159a..1e163a2bfb6ad 100644
--- a/vllm/engine/arg_utils.py
+++ b/vllm/engine/arg_utils.py
@@ -184,7 +184,7 @@ def create_engine_configs(
                                          self.worker_use_ray)
         scheduler_config = SchedulerConfig(self.max_num_batched_tokens,
                                            self.max_num_seqs,
-                                           model_config.get_max_model_len())
+                                           model_config.max_model_len)
         return model_config, cache_config, parallel_config, scheduler_config
 
 
diff --git a/vllm/engine/llm_engine.py b/vllm/engine/llm_engine.py
index 3e1026bfaefce..c1874f13f07da 100644
--- a/vllm/engine/llm_engine.py
+++ b/vllm/engine/llm_engine.py
@@ -77,6 +77,7 @@ def __init__(
             f"revision={model_config.revision}, "
             f"trust_remote_code={model_config.trust_remote_code}, "
             f"dtype={model_config.dtype}, "
+            f"max_seq_len={model_config.max_model_len}, "
             f"download_dir={model_config.download_dir!r}, "
             f"load_format={model_config.load_format}, "
             f"tensor_parallel_size={parallel_config.tensor_parallel_size}, "
diff --git a/vllm/entrypoints/openai/api_server.py b/vllm/entrypoints/openai/api_server.py
index 643dd06cb17dd..7ec155d2e488c 100644
--- a/vllm/entrypoints/openai/api_server.py
+++ b/vllm/entrypoints/openai/api_server.py
@@ -615,7 +615,7 @@ async def fake_stream_generator() -> AsyncGenerator[str, None]:
     engine_args = AsyncEngineArgs.from_cli_args(args)
     engine = AsyncLLMEngine.from_engine_args(engine_args)
     engine_model_config = asyncio.run(engine.get_model_config())
-    max_model_len = engine_model_config.get_max_model_len()
+    max_model_len = engine_model_config.max_model_len
 
     # A separate tokenizer to map token IDs to strings.
     tokenizer = get_tokenizer(engine_args.tokenizer,