diff --git a/environments/answer_format_environment/answer_format_environment.py b/environments/answer_format_environment/answer_format_environment.py
index 949f8f9e..c84d53c7 100644
--- a/environments/answer_format_environment/answer_format_environment.py
+++ b/environments/answer_format_environment/answer_format_environment.py
@@ -3653,14 +3653,6 @@ class AnswerFormatEnv(BaseEnv):
                     )
                 await self._save_failed_rollouts_to_jsonl()
 
-        # Check if all scores are the same (no learning signal)
-        if all(group_scores[0] == score for score in group_scores):
-            if self.debug_logging:
-                self.logger.debug(
-                    "All scores are identical, returning None for learning signal"
-                )
-            return None
-
         # Track successful groups for equivalent ratio enforcement
         if self.ensure_equivalent_ratios:
             # Count this as a successful group if we have any successful examples
diff --git a/environments/community/cat_behavior_env/catbot_arena.py b/environments/community/cat_behavior_env/catbot_arena.py
index 1b2d69fd..8de5a843 100644
--- a/environments/community/cat_behavior_env/catbot_arena.py
+++ b/environments/community/cat_behavior_env/catbot_arena.py
@@ -285,8 +285,6 @@ class GSM8kEnv(BaseEnv):
                         percentage_of_range = min(percentage_of_range, 1.0)
                         # Apply linear penalty scaling from 1.0 down to 0.0
                         scores["scores"].append(1.0 - percentage_of_range)
-            if all([scores["scores"][0] == score for score in scores["scores"]]):
-                return None  # If all the same, we return None
             return scores
         else:
             # If the gold solution is not parseable, we return None
diff --git a/environments/community/dynastai/dynastai_server.py b/environments/community/dynastai/dynastai_server.py
index 1ea3af05..0678e96e 100755
--- a/environments/community/dynastai/dynastai_server.py
+++ b/environments/community/dynastai/dynastai_server.py
@@ -503,11 +503,6 @@ class DynastAIEnv(BaseEnv):
         for score in scores["scores"]:
             self.percent_correct_buffer.append(max(score, 0))
 
-        # Check if all the same
-        if all([score == scores["scores"][0] for score in scores["scores"]]):
-            print("[DYNASTAI] All scores identical, returning None")
-            return None  # If all the same, we return None
-
         return scores
 
     async def get_next_item(self) -> DynastAIRow:
diff --git a/environments/community/mcp_tool_calling/tool_calling_server.py b/environments/community/mcp_tool_calling/tool_calling_server.py
index 3c765c72..57d2cd56 100644
--- a/environments/community/mcp_tool_calling/tool_calling_server.py
+++ b/environments/community/mcp_tool_calling/tool_calling_server.py
@@ -404,10 +404,6 @@ class SingleToolCallingEnv(BaseEnv):
                     # Apply linear penalty scaling from 1.0 down to 0.0
                     scores["scores"].append(1.0 - percentage_of_range)
 
-        # Check if all scores are the same (no learning signal)
-        if all(scores["scores"][0] == score for score in scores["scores"]):
-            return None
-
         return scores
 
     async def get_next_item(self):
diff --git a/environments/community/options_iv_prediction/options_iv_prediction.py b/environments/community/options_iv_prediction/options_iv_prediction.py
index 16d7b9fe..d771516e 100644
--- a/environments/community/options_iv_prediction/options_iv_prediction.py
+++ b/environments/community/options_iv_prediction/options_iv_prediction.py
@@ -433,10 +433,6 @@ class OptionsIVPrediction(BaseEnv):
             if len(scores["tokens"]) >= self.config.group_size:
                 break
 
-        # Return None if all scores are the same (no learning signal)
-        if all(scores["scores"][0] == score for score in scores["scores"]):
-            return None
-
         return scores
 
     async def rollout_and_score_eval(self, test_item):
diff --git a/environments/community/pay_to_play/pay_to_play_env.py b/environments/community/pay_to_play/pay_to_play_env.py
index 1eacdb8b..d9760d41 100644
--- a/environments/community/pay_to_play/pay_to_play_env.py
+++ b/environments/community/pay_to_play/pay_to_play_env.py
@@ -777,10 +777,6 @@ End your evaluation with \\boxed{{score}} where score is your numerical rating.
         if hasattr(self, "last_agent_card_feedback"):
             self.last_agent_card_feedback = agent_card_feedback
 
-        # Ensure we have different scores for training signal
-        if len(set(scores["scores"])) == 1:
-            return None
-
         return scores
 
     def _extract_score_from_agent_card(self, agent_card_response: str) -> float:
diff --git a/environments/community/physical_space_stl/physical_env.py b/environments/community/physical_space_stl/physical_env.py
index 7cedb2ea..91c56672 100644
--- a/environments/community/physical_space_stl/physical_env.py
+++ b/environments/community/physical_space_stl/physical_env.py
@@ -409,9 +409,6 @@ class PhysicalEnv(BaseEnv):
                         percentage_of_range = min(percentage_of_range, 1.0)
                         scores["scores"].append(1.0 - percentage_of_range)
 
-        if all([scores["scores"][0] == score for score in scores["scores"]]):
-            return None  # If all the same, we return None
-
         return scores
 
     async def get_next_item(self) -> PhysicalRow:
diff --git a/environments/community/regex_generation/regex_env.py b/environments/community/regex_generation/regex_env.py
index baa16e59..30fe311b 100644
--- a/environments/community/regex_generation/regex_env.py
+++ b/environments/community/regex_generation/regex_env.py
@@ -248,10 +248,6 @@ class RegexEnv(BaseEnv):
                 1.0 if s >= self.config.score_threshold else 0.0
             )
 
-        # If all scores identical, no learning signal
-        if len(set(scores["scores"])) == 1:
-            return None
-
         return scores
 
     async def rollout_and_score_eval(self, problem: dict) -> dict:
diff --git a/environments/community/solitaire_winning_probability/solitaire_server.py b/environments/community/solitaire_winning_probability/solitaire_server.py
index e38f02f8..c106d067 100644
--- a/environments/community/solitaire_winning_probability/solitaire_server.py
+++ b/environments/community/solitaire_winning_probability/solitaire_server.py
@@ -377,8 +377,6 @@ class SolitaireEnv(BaseEnv):
                         percentage_of_range = min(percentage_of_range, 1.0)
                         # Apply linear penalty scaling from 1.0 down to 0.0
                         scores["scores"].append(1.0 - percentage_of_range)
-            if all([scores["scores"][0] == score for score in scores["scores"]]):
-                return None  # If all the same, we return None
             return scores
         else:
             # If the gold solution is not parseable, we return None
diff --git a/environments/community/sql_query_env/sql_query_env.py b/environments/community/sql_query_env/sql_query_env.py
index 1e2a75f2..8f86e4c8 100644
--- a/environments/community/sql_query_env/sql_query_env.py
+++ b/environments/community/sql_query_env/sql_query_env.py
@@ -408,9 +408,6 @@ class SQLQueryEnv(BaseEnv):
                     percentage_of_range = min(percentage_of_range, 1.0)
                     scores["scores"].append(1.0 - percentage_of_range)
 
-        if all([scores["scores"][0] == score for score in scores["scores"]]):
-            return None  # If all the same, return None
-
         return scores
 
     async def get_next_item(self) -> WikiSQLRow:
diff --git a/environments/community/wikipedia_research/tool_calling_server.py b/environments/community/wikipedia_research/tool_calling_server.py
index e14bc717..e51cba19 100644
--- a/environments/community/wikipedia_research/tool_calling_server.py
+++ b/environments/community/wikipedia_research/tool_calling_server.py
@@ -404,10 +404,6 @@ class SingleToolCallingEnv(BaseEnv):
                     # Apply linear penalty scaling from 1.0 down to 0.0
                     scores["scores"].append(1.0 - percentage_of_range)
 
-        # Check if all scores are the same (no learning signal)
-        if all(scores["scores"][0] == score for score in scores["scores"]):
-            return None
-
         return scores
 
     async def get_next_item(self):
diff --git a/environments/eval_environments/pairwise_judgement_environment.py b/environments/eval_environments/pairwise_judgement_environment.py
index 06afae54..6fc894b3 100644
--- a/environments/eval_environments/pairwise_judgement_environment.py
+++ b/environments/eval_environments/pairwise_judgement_environment.py
@@ -1112,10 +1112,6 @@ class PairwiseJudgementEnv(BaseEnv):
             for score in scores["scores"]:
                 self.percent_correct_buffer.append(max(score, 0))
 
-            # Return None if all scores are the same (no learning signal)
-            if len(set(scores["scores"])) == 1:
-                return None
-
             return scores
 
         except Exception as e:
diff --git a/environments/fundamental_prediction_environment.py b/environments/fundamental_prediction_environment.py
index 9c359306..eeaaed98 100644
--- a/environments/fundamental_prediction_environment.py
+++ b/environments/fundamental_prediction_environment.py
@@ -402,10 +402,6 @@ class FundamentalPredictionEnv(BaseEnv):
             if len(scores["tokens"]) >= self.config.group_size:
                 break
 
-        # Return None if all scores are the same (no learning signal)
-        if all(scores["scores"][0] == score for score in scores["scores"]):
-            return None
-
         return scores
 
     async def rollout_and_score_eval(self, test_item):
diff --git a/environments/gsm8k_server.py b/environments/gsm8k_server.py
index 6ae5285b..8b3589da 100644
--- a/environments/gsm8k_server.py
+++ b/environments/gsm8k_server.py
@@ -352,8 +352,6 @@ class GSM8kEnv(BaseEnv):
                         percentage_of_range = min(percentage_of_range, 1.0)
                         # Apply linear penalty scaling from 1.0 down to 0.0
                         scores["scores"].append(1.0 - percentage_of_range)
-            if all([scores["scores"][0] == score for score in scores["scores"]]):
-                return None  # If all the same, we return None
             return scores
         else:
             # If the gold solution is not parseable, we return None
diff --git a/environments/gsm8k_server_axolotl.py b/environments/gsm8k_server_axolotl.py
index 244861f8..571eb609 100644
--- a/environments/gsm8k_server_axolotl.py
+++ b/environments/gsm8k_server_axolotl.py
@@ -283,8 +283,6 @@ class GSM8kEnv(BaseEnv):
                         percentage_of_range = min(percentage_of_range, 1.0)
                         # Apply linear penalty scaling from 1.0 down to 0.0
                         scores["scores"].append(1.0 - percentage_of_range)
-            if all([scores["scores"][0] == score for score in scores["scores"]]):
-                return None  # If all the same, we return None
             return scores
         else:
             # If the gold solution is not parseable, we return None
diff --git a/environments/letter_counting_environment/letter_counting_environment.py b/environments/letter_counting_environment/letter_counting_environment.py
index 2398373b..6f4f9cf1 100644
--- a/environments/letter_counting_environment/letter_counting_environment.py
+++ b/environments/letter_counting_environment/letter_counting_environment.py
@@ -1248,13 +1248,6 @@ class LetterCountingEnv(BaseEnv):
             )
             return None
 
-        # Skip if all scores are identical (no learning signal)
-        if all(scores["scores"][0] == score for score in scores["scores"]):
-            self.logger.debug(
-                f"All scores identical ({scores['scores'][0]:.2f}) - skipping group"
-            )
-            return None
-
         return scores
 
     async def rollout_and_score_eval(self, eval_item: Dict) -> Tuple[int, int]:
diff --git a/environments/math_server.py b/environments/math_server.py
index 5b343b9f..b0dafbfd 100644
--- a/environments/math_server.py
+++ b/environments/math_server.py
@@ -534,8 +534,6 @@ class MathEnv(BaseEnv):
                         return None, to_backlog
                 else:
                     return None, to_backlog
-            else:
-                return None, to_backlog
         else:
             self.normal_rollouts.append(
                 (
@@ -1167,8 +1165,6 @@ class MathEnv(BaseEnv):
                     "Max message delta is less than 0.1 * shortest message, no length penalty"
                 )
                 return None, []
-        elif all([score == scores["scores"][0] for score in scores["scores"]]):
-            return None, []
         if len(for_table) > 0:
             self.judge_rollouts.append(for_table)
             if len(self.judge_rollouts) >= self.config.num_rollouts_to_keep:
diff --git a/environments/math_server_zero.py b/environments/math_server_zero.py
index 1432ab4d..43cd6a09 100644
--- a/environments/math_server_zero.py
+++ b/environments/math_server_zero.py
@@ -383,10 +383,6 @@ class MathEnv(BaseEnv):
         to_postprocess = await self.score(to_score)
         if to_postprocess is None:
             return None, to_backlog
-        if all(
-            [to_postprocess["scores"][0] == score for score in to_postprocess["scores"]]
-        ):
-            return None, to_backlog
         self.normal_rollouts.append(
             (
                 prompt_format.format(prompt=problem_format.format(problem=item[0])),
diff --git a/environments/mcp_env.py b/environments/mcp_env.py
index a8d621bc..f8892cb1 100644
--- a/environments/mcp_env.py
+++ b/environments/mcp_env.py
@@ -347,9 +347,6 @@ class McpEnv(BaseEnv):
         for score in scores["scores"]:
             self.percent_correct_buffer.append(max(score, 0))
 
-        if all(scores["scores"][0] == score for score in scores["scores"]):
-            return None
-
         return scores
 
     async def get_next_item(self):
diff --git a/environments/mcqa_thinking_env.py b/environments/mcqa_thinking_env.py
index 417822f6..9738a85e 100644
--- a/environments/mcqa_thinking_env.py
+++ b/environments/mcqa_thinking_env.py
@@ -373,10 +373,6 @@ class MCQAThinkingEnv(BaseEnv):
         for score in scores["scores"]:
             self.percent_correct_buffer.append(max(score, 0))
 
-        # Return None if all scores are the same (no learning signal)
-        if all(scores["scores"][0] == score for score in scores["scores"]):
-            return None
-
         return scores
 
     async def rollout_and_score_eval(self, test_item):
diff --git a/environments/text_reversal_environment.py b/environments/text_reversal_environment.py
index 0b41e2ec..738cb663 100644
--- a/environments/text_reversal_environment.py
+++ b/environments/text_reversal_environment.py
@@ -770,10 +770,6 @@ class TextReversalEnv(BaseEnv):
             for score in scores["scores"]:
                 self.percent_correct_buffer.append(max(score, 0))
 
-            # Return None if all scores are the same (no learning signal)
-            if len(set(scores["scores"])) == 1:
-                return None
-
             return scores
 
         except Exception as e:
diff --git a/environments/tool_calling_server.py b/environments/tool_calling_server.py
index 10409417..e6824118 100644
--- a/environments/tool_calling_server.py
+++ b/environments/tool_calling_server.py
@@ -413,10 +413,6 @@ class SingleToolCallingEnv(BaseEnv):
                     # Apply linear penalty scaling from 1.0 down to 0.0
                     scores["scores"].append(1.0 - percentage_of_range)
 
-        # Check if all scores are the same (no learning signal)
-        if all(scores["scores"][0] == score for score in scores["scores"]):
-            return None
-
         return scores
 
     async def get_next_item(self):