diff --git a/studio/backend/tests/test_dataset_preference_format.py b/studio/backend/tests/test_dataset_preference_format.py
new file mode 100644
index 0000000000..36ba927491
--- /dev/null
+++ b/studio/backend/tests/test_dataset_preference_format.py
@@ -0,0 +1,87 @@
+# SPDX-License-Identifier: AGPL-3.0-only
+# Copyright 2026-present the Unsloth AI Inc. team. All rights reserved. See /studio/LICENSE.AGPL-3.0
+
+from datasets import Dataset
+
+from utils.datasets.dataset_utils import (
+    check_dataset_format,
+    format_and_template_dataset,
+    format_dataset,
+)
+
+
+def _preference_dataset():
+    return Dataset.from_list(
+        [
+            {
+                "prompt": "The sky is",
+                "chosen": " blue.",
+                "rejected": " green.",
+            }
+        ]
+    )
+
+
+def test_check_dataset_format_detects_preference_dataset():
+    result = check_dataset_format(_preference_dataset())
+
+    assert result["detected_format"] == "preference"
+    assert result["requires_manual_mapping"] is False
+
+
+def test_check_dataset_format_requires_review_when_prompt_is_missing():
+    dataset = Dataset.from_list([{"chosen": " blue.", "rejected": " green."}])
+
+    result = check_dataset_format(dataset)
+
+    assert result["detected_format"] == "unknown"
+    assert result["requires_manual_mapping"] is True
+
+
+def test_format_dataset_auto_preserves_preference_dataset():
+    result = format_dataset(_preference_dataset(), format_type = "auto")
+
+    assert result["detected_format"] == "preference"
+    assert result["final_format"] == "preference"
+    assert result["requires_manual_mapping"] is False
+    assert result["dataset"][0]["chosen"] == " blue."
+
+
+def test_format_dataset_preference_mapping_converts_custom_columns():
+    dataset = Dataset.from_list(
+        [{"question": "The sky is", "winner": " blue.", "loser": " green."}]
+    )
+
+    result = format_dataset(
+        dataset,
+        format_type = "preference",
+        custom_format_mapping = {
+            "question": "prompt",
+            "winner": "chosen",
+            "loser": "rejected",
+        },
+    )
+
+    assert result["final_format"] == "preference"
+    assert result["dataset"][0] == {
+        "prompt": "The sky is",
+        "chosen": " blue.",
+        "rejected": " green.",
+    }
+
+
+def test_format_and_template_dataset_skips_chat_template_for_preference():
+    result = format_and_template_dataset(
+        _preference_dataset(),
+        model_name = "unsloth/Qwen2.5-0.5B-Instruct",
+        tokenizer = object(),
+        format_type = "auto",
+    )
+
+    assert result["success"] is True
+    assert result["final_format"] == "preference"
+    assert result["dataset"][0] == {
+        "prompt": "The sky is",
+        "chosen": " blue.",
+        "rejected": " green.",
+    }