diff --git a/py/autoevals/ragas.py b/py/autoevals/ragas.py
index 4726159..911d81f 100644
--- a/py/autoevals/ragas.py
+++ b/py/autoevals/ragas.py
@@ -1127,7 +1127,8 @@ class AnswerRelevancy(OpenAILLMScorer):
         output: The generated answer to evaluate
         context: The context document(s) to evaluate against
         strictness: Optional float between 0-1, higher values enforce stricter relevancy
-        temperature: Optional float between 0-1, lower values make evaluation more deterministic
+        temperature: Optional float between 0-1, lower values make evaluation more deterministic.
+            If not specified, uses the model's default
         embedding_model: Optional model to use for embeddings, defaults to text-embedding-3-small
     """
 
@@ -1135,7 +1136,7 @@ def __init__(
         self,
         model: str | None = None,
         strictness=3,
-        temperature=0.5,
+        temperature=None,
         embedding_model=None,
         client: Client | None = None,
         **kwargs,
diff --git a/py/autoevals/test_ragas.py b/py/autoevals/test_ragas.py
index 74db4e5..40d022d 100644
--- a/py/autoevals/test_ragas.py
+++ b/py/autoevals/test_ragas.py
@@ -262,3 +262,18 @@ def mock_responses_api(request):
     )
 
     assert captured_embedding_model == "text-embedding-3-large"
+
+
+def test_answer_relevancy_does_not_send_a_default_temperature():
+    # Claude 4.7+ and 5.x reject a non-default temperature, so the request only carries one the caller asked for.
+    scorer = AnswerRelevancy(model="claude-opus-5-5", client=object())
+    request = ragas_module.extract_question_gen_request(
+        answer="Paris", context="France", model=scorer.model, **scorer.extra_args
+    )
+    assert "temperature" not in request
+
+    scorer = AnswerRelevancy(model="gpt-4o", temperature=0.2, client=object())
+    request = ragas_module.extract_question_gen_request(
+        answer="Paris", context="France", model=scorer.model, **scorer.extra_args
+    )
+    assert request["temperature"] == 0.2