diff --git a/agentplatform/_genai/_evals_metric_handlers.py b/agentplatform/_genai/_evals_metric_handlers.py
index f8fafb4b53..77f5db46e4 100644
--- a/agentplatform/_genai/_evals_metric_handlers.py
+++ b/agentplatform/_genai/_evals_metric_handlers.py
@@ -686,6 +686,12 @@ def _build_evaluation_instance(
             other_data_map["context"] = types.evals.InstanceData(text=eval_case.context)
         elif isinstance(eval_case.context, genai_types.Content):
             other_data_map["context"] = _content_to_instance_data(eval_case.context)
+        else:
+            other_data_map["context"] = types.evals.InstanceData(
+                contents=types.evals.InstanceDataContents(
+                    contents=_value_to_content_list(eval_case.context)
+                )
+            )
 
     # 3. Extract custom variables from LLMMetric templates
     if prompt_template:
diff --git a/tests/unit/agentplatform/genai/test_evals.py b/tests/unit/agentplatform/genai/test_evals.py
index 6feb9aba93..a6976942ba 100644
--- a/tests/unit/agentplatform/genai/test_evals.py
+++ b/tests/unit/agentplatform/genai/test_evals.py
@@ -6950,6 +6950,28 @@ def test_build_evaluation_instance_various_field_types(self):
             == "System instructions here."
         )
 
+    @pytest.mark.parametrize(
+        "context",
+        [["Passage one.", "Passage two."], {"source": "doc-1", "text": "Passage."}],
+    )
+    def test_build_evaluation_instance_list_and_dict_context(self, context):
+        eval_case = agentplatform_genai_types.EvalCase(
+            prompt=genai_types.Content(parts=[genai_types.Part(text="Hello")]),
+            responses=[
+                agentplatform_genai_types.ResponseCandidate(
+                    response=genai_types.Content(parts=[genai_types.Part(text="Hi")])
+                )
+            ],
+            context=context,
+        )
+
+        instance = _evals_metric_handlers._build_evaluation_instance(
+            eval_case, eval_case.responses[0].response
+        )
+
+        context_data = instance.other_data.map_instance["context"]
+        assert context_data.contents.contents[0].parts[0].text == json.dumps(context)
+
 
 class TestMetric:
     """Unit tests for the Metric class."""
diff --git a/vertexai/_genai/_evals_metric_handlers.py b/vertexai/_genai/_evals_metric_handlers.py
index d682460243..79b766d4d2 100644
--- a/vertexai/_genai/_evals_metric_handlers.py
+++ b/vertexai/_genai/_evals_metric_handlers.py
@@ -684,6 +684,12 @@ def _build_evaluation_instance(
             other_data_map["context"] = types.evals.InstanceData(text=eval_case.context)
         elif isinstance(eval_case.context, genai_types.Content):
             other_data_map["context"] = _content_to_instance_data(eval_case.context)
+        else:
+            other_data_map["context"] = types.evals.InstanceData(
+                contents=types.evals.InstanceDataContents(
+                    contents=_value_to_content_list(eval_case.context)
+                )
+            )
 
     # 3. Extract custom variables from LLMMetric templates
     if prompt_template: