diff --git a/packages/prime-evals/README.md b/packages/prime-evals/README.md
index 900dcff7..52ee41fb 100644
--- a/packages/prime-evals/README.md
+++ b/packages/prime-evals/README.md
@@ -154,7 +154,7 @@ samples_batch = [
         "completion": [
             {"role": "assistant", "content": f"The answer is {i * 2}."}
         ],
-        "metadata": {"batch": 1}
+        "info": {"batch": 1}
     }
     for i in range(10)
 ]
diff --git a/packages/prime-evals/src/prime_evals/models.py b/packages/prime-evals/src/prime_evals/models.py
index 16efe6b8..7d278dcf 100644
--- a/packages/prime-evals/src/prime_evals/models.py
+++ b/packages/prime-evals/src/prime_evals/models.py
@@ -83,7 +83,7 @@ class Sample(BaseModel):
     correct: Optional[bool] = None
     format_reward: Optional[float] = Field(None, alias="formatReward")
     correctness: Optional[float] = None
-    metadata: Optional[Dict[str, Any]] = None
+    info: Optional[Dict[str, Any]] = None
 
     model_config = ConfigDict(populate_by_name=True, extra="allow")
 
diff --git a/packages/prime-evals/tests/test_evals.py b/packages/prime-evals/tests/test_evals.py
index a320089b..47183782 100644
--- a/packages/prime-evals/tests/test_evals.py
+++ b/packages/prime-evals/tests/test_evals.py
@@ -111,7 +111,7 @@ def test_sample_model_with_metadata():
         "reward": 1.0,
         "answer": "18",
         "custom_field": "custom_value",  # Extra field should be allowed
-        "metadata": {"batch": 1},
+        "info": {"batch": 1},
     }
 
     sample = Sample.model_validate(data)
@@ -119,7 +119,7 @@ def test_sample_model_with_metadata():
     assert sample.example_id == 0
     assert sample.task == "gsm8k"
     assert sample.reward == 1.0
-    assert sample.metadata == {"batch": 1}
+    assert sample.info == {"batch": 1}
 
 
 def test_evals_client_context_manager():