diff --git a/openhands/llm/llm.py b/openhands/llm/llm.py
index dcfeecd9c6..91e0c79f15 100644
--- a/openhands/llm/llm.py
+++ b/openhands/llm/llm.py
@@ -163,6 +163,7 @@ class LLM(RetryMixin, DebugMixin):
             'temperature': self.config.temperature,
             'max_completion_tokens': self.config.max_output_tokens,
         }
+
         if self.config.top_k is not None:
             # openai doesn't expose top_k
             # litellm will handle it a bit differently than the openai-compatible params
@@ -486,26 +487,6 @@ class LLM(RetryMixin, DebugMixin):
                 # Safe fallback for any potentially viable model
                 self.config.max_input_tokens = 4096
 
-        if self.config.max_output_tokens is None:
-            # Safe default for any potentially viable model
-            self.config.max_output_tokens = 4096
-            if self.model_info is not None:
-                # max_output_tokens has precedence over max_tokens, if either exists.
-                # litellm has models with both, one or none of these 2 parameters!
-                if 'max_output_tokens' in self.model_info and isinstance(
-                    self.model_info['max_output_tokens'], int
-                ):
-                    self.config.max_output_tokens = self.model_info['max_output_tokens']
-                elif 'max_tokens' in self.model_info and isinstance(
-                    self.model_info['max_tokens'], int
-                ):
-                    self.config.max_output_tokens = self.model_info['max_tokens']
-            if any(
-                model in self.config.model
-                for model in ['claude-3-7-sonnet', 'claude-3.7-sonnet']
-            ):
-                self.config.max_output_tokens = 64000  # litellm set max to 128k, but that requires a header to be set
-
         # Initialize function calling capability
         # Check if model name is in our supported list
         model_name_supported = (
diff --git a/tests/unit/test_llm.py b/tests/unit/test_llm.py
index 77c24f7da1..94840bf39d 100644
--- a/tests/unit/test_llm.py
+++ b/tests/unit/test_llm.py
@@ -132,7 +132,7 @@ def test_llm_init_with_model_info(mock_get_model_info, default_config):
     llm = LLM(default_config)
     llm.init_model_info()
     assert llm.config.max_input_tokens == 8000
-    assert llm.config.max_output_tokens == 2000
+    assert llm.config.max_output_tokens is None
 
 
 @patch('openhands.llm.llm.litellm.get_model_info')
@@ -141,7 +141,7 @@ def test_llm_init_without_model_info(mock_get_model_info, default_config):
     llm = LLM(default_config)
     llm.init_model_info()
     assert llm.config.max_input_tokens == 4096
-    assert llm.config.max_output_tokens == 4096
+    assert llm.config.max_output_tokens is None
 
 
 def test_llm_init_with_custom_config():
@@ -260,7 +260,7 @@ def test_llm_init_with_openrouter_model(mock_get_model_info, default_config):
     llm = LLM(default_config)
     llm.init_model_info()
     assert llm.config.max_input_tokens == 7000
-    assert llm.config.max_output_tokens == 1500
+    assert llm.config.max_output_tokens is None
     mock_get_model_info.assert_called_once_with('openrouter:gpt-4o-mini')