diff --git a/tests/test_litellm/proxy/db/test_litellm_config_cache.py b/tests/test_litellm/proxy/db/test_litellm_config_cache.py
new file mode 100644
index 00000000000..2ddbc5b0681
--- /dev/null
+++ b/tests/test_litellm/proxy/db/test_litellm_config_cache.py
@@ -0,0 +1,94 @@
+"""
+Tests for batch_load_config - batching LiteLLM_Config queries into one find_many.
+"""
+
+from types import SimpleNamespace
+from unittest.mock import AsyncMock, MagicMock
+
+import pytest
+
+from litellm.proxy.db.litellm_config_cache import (
+    POLLING_PARAM_NAMES,
+    batch_load_config,
+)
+
+
+def _make_config_record(param_name: str, param_value: dict):
+    """Helper to create a mock LiteLLM_Config record."""
+    return SimpleNamespace(param_name=param_name, param_value=param_value)
+
+
+def _make_mock_prisma_client(records):
+    """Helper to create a mock PrismaClient with find_many returning records."""
+    mock_client = MagicMock()
+    mock_client.db.litellm_config.find_many = AsyncMock(return_value=records)
+    return mock_client
+
+
+@pytest.mark.asyncio
+async def test_batch_load_fetches_all_config_in_one_query():
+    """batch_load_config should call find_many once with all param names."""
+    records = [
+        _make_config_record("general_settings", {"key": "value"}),
+        _make_config_record("litellm_settings", {"setting": True}),
+    ]
+    mock_client = _make_mock_prisma_client(records)
+
+    result = await batch_load_config(mock_client)
+
+    mock_client.db.litellm_config.find_many.assert_called_once_with(
+        where={"param_name": {"in": POLLING_PARAM_NAMES}}
+    )
+    assert len(result) == 2
+
+
+@pytest.mark.asyncio
+async def test_batch_load_returns_dict_keyed_by_param_name():
+    """Result should be a dict keyed by param_name."""
+    records = [
+        _make_config_record("general_settings", {"alerting": ["slack"]}),
+        _make_config_record("model_cost_map_reload_config", {"interval_hours": 24}),
+    ]
+    mock_client = _make_mock_prisma_client(records)
+
+    result = await batch_load_config(mock_client)
+
+    assert result["general_settings"].param_value == {"alerting": ["slack"]}
+    assert result["model_cost_map_reload_config"].param_value == {"interval_hours": 24}
+
+
+@pytest.mark.asyncio
+async def test_batch_load_returns_empty_dict_for_no_records():
+    """If no config records exist, should return empty dict."""
+    mock_client = _make_mock_prisma_client([])
+
+    result = await batch_load_config(mock_client)
+
+    assert result == {}
+
+
+@pytest.mark.asyncio
+async def test_batch_load_missing_key_returns_none_on_get():
+    """dict.get() on a missing key should return None."""
+    records = [
+        _make_config_record("general_settings", {"key": "value"}),
+    ]
+    mock_client = _make_mock_prisma_client(records)
+
+    result = await batch_load_config(mock_client)
+
+    assert result.get("nonexistent_key") is None
+    assert result.get("general_settings") is not None
+
+
+@pytest.mark.asyncio
+async def test_batch_load_contains_all_polling_param_names_when_all_exist():
+    """When all param_names exist in DB, all should be in the result."""
+    records = [_make_config_record(name, {"data": name}) for name in POLLING_PARAM_NAMES]
+    mock_client = _make_mock_prisma_client(records)
+
+    result = await batch_load_config(mock_client)
+
+    for name in POLLING_PARAM_NAMES:
+        assert name in result
+        assert result[name].param_value == {"data": name}