diff --git a/nbs/distributed.timegpt.ipynb b/nbs/distributed.timegpt.ipynb
index 3d9774ab..87e51901 100644
--- a/nbs/distributed.timegpt.ipynb
+++ b/nbs/distributed.timegpt.ipynb
@@ -161,6 +161,7 @@
     "            X_df: Optional[fugue.AnyDataFrame] = None,\n",
     "            level: Optional[List[Union[int, float]]] = None,\n",
     "            finetune_steps: int = 0,\n",
+    "            finetune_loss: str = 'default',\n",
     "            clean_ex_first: bool = True,\n",
     "            validate_token: bool = False,\n",
     "            add_history: bool = False,\n",
@@ -177,6 +178,7 @@
     "            target_col=target_col,\n",
     "            level=level,\n",
     "            finetune_steps=finetune_steps,\n",
+    "            finetune_loss=finetune_loss,\n",
     "            clean_ex_first=clean_ex_first,\n",
     "            validate_token=validate_token,\n",
     "            add_history=add_history,\n",
@@ -245,6 +247,7 @@
     "            target_col: str = 'y',\n",
     "            level: Optional[List[Union[int, float]]] = None,\n",
     "            finetune_steps: int = 0,\n",
+    "            finetune_loss: str = 'default',\n",
     "            clean_ex_first: bool = True,\n",
     "            validate_token: bool = False,\n",
     "            date_features: Union[bool, List[str]] = False,\n",
@@ -262,6 +265,7 @@
     "            target_col=target_col,\n",
     "            level=level,\n",
     "            finetune_steps=finetune_steps,\n",
+    "            finetune_loss=finetune_loss,\n",
     "            clean_ex_first=clean_ex_first,\n",
     "            validate_token=validate_token,\n",
     "            date_features=date_features,\n",
diff --git a/nbs/timegpt.ipynb b/nbs/timegpt.ipynb
index ee172ed4..b67a89b7 100644
--- a/nbs/timegpt.ipynb
+++ b/nbs/timegpt.ipynb
@@ -164,6 +164,7 @@
     "            freq: str = None,\n",
     "            level: Optional[List[Union[int, float]]] = None,\n",
     "            finetune_steps: int = 0,\n",
+    "            finetune_loss: str = 'default',\n",
     "            clean_ex_first: bool = True,\n",
     "            date_features: Union[bool, List[str]] = False,\n",
     "            date_features_to_one_hot: Union[bool, List[str]] = True,\n",
@@ -180,6 +181,7 @@
     "        self.base_freq = freq\n",
     "        self.level = level\n",
     "        self.finetune_steps = finetune_steps\n",
+    "        self.finetune_loss = finetune_loss\n",
     "        self.clean_ex_first = clean_ex_first\n",
     "        self.date_features = date_features\n",
     "        self.date_features_to_one_hot = date_features_to_one_hot\n",
@@ -510,6 +512,7 @@
     "            freq=self.freq,\n",
     "            level=self.level,\n",
     "            finetune_steps=self.finetune_steps,\n",
+    "            finetune_loss=self.finetune_loss,\n",
     "            clean_ex_first=self.clean_ex_first,\n",
     "            model=self.model,\n",
     "        )\n",
@@ -798,6 +801,7 @@
     "            X_df: Optional[pd.DataFrame] = None,\n",
     "            level: Optional[List[Union[int, float]]] = None,\n",
     "            finetune_steps: int = 0,\n",
+    "            finetune_loss: str = 'default',\n",
     "            clean_ex_first: bool = True,\n",
     "            validate_token: bool = False,\n",
     "            add_history: bool = False,\n",
@@ -819,6 +823,7 @@
     "            freq=freq,\n",
     "            level=level,\n",
     "            finetune_steps=finetune_steps,\n",
+    "            finetune_loss=finetune_loss,\n",
     "            clean_ex_first=clean_ex_first,\n",
     "            date_features=date_features,\n",
     "            date_features_to_one_hot=date_features_to_one_hot,\n",
@@ -887,6 +892,7 @@
     "        n_windows: int = 1,\n",
     "        step_size: Optional[int] = None,\n",
     "        finetune_steps: int = 0,\n",
+    "        finetune_loss: str = 'default',\n",
     "        clean_ex_first: bool = True,\n",
     "        date_features: Union[bool, List[str]] = False,\n",
     "        date_features_to_one_hot: Union[bool, List[str]] = True,\n",
@@ -906,6 +912,7 @@
     "            freq=freq,\n",
     "            level=level,\n",
     "            finetune_steps=finetune_steps,\n",
+    "            finetune_loss=finetune_loss,\n",
     "            clean_ex_first=clean_ex_first,\n",
     "            date_features=date_features,\n",
     "            date_features_to_one_hot=date_features_to_one_hot,\n",
@@ -1059,6 +1066,7 @@
     "            X_df: Optional[pd.DataFrame] = None,\n",
     "            level: Optional[List[Union[int, float]]] = None,\n",
     "            finetune_steps: int = 0,\n",
+    "            finetune_loss: str = 'default',\n",
     "            clean_ex_first: bool = True,\n",
     "            validate_token: bool = False,\n",
     "            add_history: bool = False,\n",
@@ -1101,6 +1109,8 @@
     "        finetune_steps : int (default=0)\n",
     "            Number of steps used to finetune TimeGPT in the\n",
     "            new data.\n",
+    "        finetune_loss : str (default='default')\n",
+    "            Loss function to use for finetuning. Options are: `default`, `mae`, `mse`, `rmse`, `mape`, and `smape`.\n",
     "        clean_ex_first : bool (default=True)\n",
     "            Clean exogenous signal before making forecasts\n",
     "            using TimeGPT.\n",
@@ -1145,6 +1155,7 @@
     "                X_df=X_df,\n",
     "                level=level,\n",
     "                finetune_steps=finetune_steps,\n",
+    "                finetune_loss=finetune_loss,\n",
     "                clean_ex_first=clean_ex_first,\n",
     "                validate_token=validate_token,\n",
     "                add_history=add_history,\n",
@@ -1165,6 +1176,7 @@
     "                X_df=X_df,\n",
     "                level=level,\n",
     "                finetune_steps=finetune_steps,\n",
+    "                finetune_loss=finetune_loss,\n",
     "                clean_ex_first=clean_ex_first,\n",
     "                validate_token=validate_token,\n",
     "                add_history=add_history,\n",
@@ -1291,6 +1303,7 @@
     "        n_windows: int = 1,\n",
     "        step_size: Optional[int] = None,\n",
     "        finetune_steps: int = 0,\n",
+    "        finetune_loss: str = 'default',\n",
     "        clean_ex_first: bool = True,\n",
     "        date_features: Union[bool, List[str]] = False,\n",
     "        date_features_to_one_hot: Union[bool, List[str]] = True,\n",
@@ -1336,6 +1349,8 @@
     "        finetune_steps : int (default=0)\n",
     "            Number of steps used to finetune TimeGPT in the\n",
     "            new data.\n",
+    "        finetune_loss : str (default='default')\n",
+    "            Loss function to use for finetuning. Options are: `default`, `mae`, `mse`, `rmse`, `mape`, and `smape`.\n",
     "        clean_ex_first : bool (default=True)\n",
     "            Clean exogenous signal before making forecasts\n",
     "            using TimeGPT.\n",
@@ -1373,6 +1388,7 @@
     "                target_col=target_col,\n",
     "                level=level,\n",
     "                finetune_steps=finetune_steps,\n",
+    "                finetune_loss=finetune_loss,\n",
     "                clean_ex_first=clean_ex_first,\n",
     "                validate_token=validate_token,\n",
     "                date_features=date_features,\n",
@@ -1393,6 +1409,7 @@
     "                target_col=target_col,\n",
     "                level=level,\n",
     "                finetune_steps=finetune_steps,\n",
+    "                finetune_loss=finetune_loss,\n",
     "                clean_ex_first=clean_ex_first,\n",
     "                validate_token=validate_token,\n",
     "                date_features=date_features,\n",
@@ -1862,7 +1879,7 @@
     "#| hide\n",
     "#test same results custom url\n",
     "timegpt_custom = TimeGPT(\n",
-    "    token=os.environ['TIMEGPT_CUSTOM_URL_TOKEN'],\n",
+    "    token=os.environ['TIMEGPT_CUSTOM_URL_TOKEN'], \n",
     "    environment=os.environ['TIMEGPT_CUSTOM_URL'],\n",
     ")\n",
     "# forecast method\n",
@@ -2294,7 +2311,7 @@
     "#| hide \n",
     "# test short horizon with finetunning\n",
     "test_fail(\n",
-    "    lambda: timegpt.forecast(df=df.tail(3), h=12, time_col='timestamp', target_col='value', finetune_steps=10),\n",
+    "    lambda: timegpt.forecast(df=df.tail(3), h=12, time_col='timestamp', target_col='value', finetune_steps=10, finetune_loss='mae'),\n",
     "    contains='be sure'\n",
     ")"
    ]
diff --git a/nixtlats/_modidx.py b/nixtlats/_modidx.py
index 92dd7362..c91e07ed 100644
--- a/nixtlats/_modidx.py
+++ b/nixtlats/_modidx.py
@@ -119,8 +119,17 @@
                                   'nixtlats.timegpt.validate_model_parameter': ( 'timegpt.html#validate_model_parameter',
                                                                                  'nixtlats/timegpt.py')},
             'nixtlats.types.http_validation_error': {},
+            'nixtlats.types.multi_series_anomaly_model': {},
+            'nixtlats.types.multi_series_cross_validation_finetune_loss': {},
+            'nixtlats.types.multi_series_cross_validation_model': {},
+            'nixtlats.types.multi_series_forecast_finetune_loss': {},
+            'nixtlats.types.multi_series_forecast_model': {},
             'nixtlats.types.multi_series_input': {},
+            'nixtlats.types.multi_series_insample_forecast_model': {},
             'nixtlats.types.single_series_forecast': {},
+            'nixtlats.types.single_series_forecast_finetune_loss': {},
+            'nixtlats.types.single_series_forecast_model': {},
+            'nixtlats.types.single_series_insample_forecast_model': {},
             'nixtlats.types.validation_error': {},
             'nixtlats.types.validation_error_loc_item': {},
             'nixtlats.utils': {'nixtlats.utils.colab_badge': ('utils.html#colab_badge', 'nixtlats/utils.py')}}}
diff --git a/nixtlats/distributed/timegpt.py b/nixtlats/distributed/timegpt.py
index be110a89..9fb9a243 100644
--- a/nixtlats/distributed/timegpt.py
+++ b/nixtlats/distributed/timegpt.py
@@ -125,6 +125,7 @@ def forecast(
         X_df: Optional[fugue.AnyDataFrame] = None,
         level: Optional[List[Union[int, float]]] = None,
         finetune_steps: int = 0,
+        finetune_loss: str = "default",
         clean_ex_first: bool = True,
         validate_token: bool = False,
         add_history: bool = False,
@@ -141,6 +142,7 @@ def forecast(
             target_col=target_col,
             level=level,
             finetune_steps=finetune_steps,
+            finetune_loss=finetune_loss,
             clean_ex_first=clean_ex_first,
             validate_token=validate_token,
             add_history=add_history,
@@ -211,6 +213,7 @@ def cross_validation(
         target_col: str = "y",
         level: Optional[List[Union[int, float]]] = None,
         finetune_steps: int = 0,
+        finetune_loss: str = "default",
         clean_ex_first: bool = True,
         validate_token: bool = False,
         date_features: Union[bool, List[str]] = False,
@@ -228,6 +231,7 @@ def cross_validation(
             target_col=target_col,
             level=level,
             finetune_steps=finetune_steps,
+            finetune_loss=finetune_loss,
             clean_ex_first=clean_ex_first,
             validate_token=validate_token,
             date_features=date_features,
diff --git a/nixtlats/timegpt.py b/nixtlats/timegpt.py
index cbeb4a73..326d9f9a 100644
--- a/nixtlats/timegpt.py
+++ b/nixtlats/timegpt.py
@@ -100,6 +100,7 @@ def __init__(
         freq: str = None,
         level: Optional[List[Union[int, float]]] = None,
         finetune_steps: int = 0,
+        finetune_loss: str = "default",
         clean_ex_first: bool = True,
         date_features: Union[bool, List[str]] = False,
         date_features_to_one_hot: Union[bool, List[str]] = True,
@@ -116,6 +117,7 @@ def __init__(
         self.base_freq = freq
         self.level = level
         self.finetune_steps = finetune_steps
+        self.finetune_loss = finetune_loss
         self.clean_ex_first = clean_ex_first
         self.date_features = date_features
         self.date_features_to_one_hot = date_features_to_one_hot
@@ -481,6 +483,7 @@ def forecast(
             freq=self.freq,
             level=self.level,
             finetune_steps=self.finetune_steps,
+            finetune_loss=self.finetune_loss,
             clean_ex_first=self.clean_ex_first,
             model=self.model,
         )
@@ -749,6 +752,7 @@ def _forecast(
         X_df: Optional[pd.DataFrame] = None,
         level: Optional[List[Union[int, float]]] = None,
         finetune_steps: int = 0,
+        finetune_loss: str = "default",
         clean_ex_first: bool = True,
         validate_token: bool = False,
         add_history: bool = False,
@@ -768,6 +772,7 @@ def _forecast(
             freq=freq,
             level=level,
             finetune_steps=finetune_steps,
+            finetune_loss=finetune_loss,
             clean_ex_first=clean_ex_first,
             date_features=date_features,
             date_features_to_one_hot=date_features_to_one_hot,
@@ -834,6 +839,7 @@ def _cross_validation(
         n_windows: int = 1,
         step_size: Optional[int] = None,
         finetune_steps: int = 0,
+        finetune_loss: str = "default",
         clean_ex_first: bool = True,
         date_features: Union[bool, List[str]] = False,
         date_features_to_one_hot: Union[bool, List[str]] = True,
@@ -851,6 +857,7 @@ def _cross_validation(
             freq=freq,
             level=level,
             finetune_steps=finetune_steps,
+            finetune_loss=finetune_loss,
             clean_ex_first=clean_ex_first,
             date_features=date_features,
             date_features_to_one_hot=date_features_to_one_hot,
@@ -999,6 +1006,7 @@ def forecast(
         X_df: Optional[pd.DataFrame] = None,
         level: Optional[List[Union[int, float]]] = None,
         finetune_steps: int = 0,
+        finetune_loss: str = "default",
         clean_ex_first: bool = True,
         validate_token: bool = False,
         add_history: bool = False,
@@ -1041,6 +1049,8 @@ def forecast(
         finetune_steps : int (default=0)
             Number of steps used to finetune TimeGPT in the
             new data.
+        finetune_loss : str (default='default')
+            Loss function to use for finetuning. Options are: `default`, `mae`, `mse`, `rmse`, `mape`, and `smape`.
         clean_ex_first : bool (default=True)
             Clean exogenous signal before making forecasts
             using TimeGPT.
@@ -1085,6 +1095,7 @@ def forecast(
                 X_df=X_df,
                 level=level,
                 finetune_steps=finetune_steps,
+                finetune_loss=finetune_loss,
                 clean_ex_first=clean_ex_first,
                 validate_token=validate_token,
                 add_history=add_history,
@@ -1105,6 +1116,7 @@ def forecast(
                 X_df=X_df,
                 level=level,
                 finetune_steps=finetune_steps,
+                finetune_loss=finetune_loss,
                 clean_ex_first=clean_ex_first,
                 validate_token=validate_token,
                 add_history=add_history,
@@ -1231,6 +1243,7 @@ def cross_validation(
         n_windows: int = 1,
         step_size: Optional[int] = None,
         finetune_steps: int = 0,
+        finetune_loss: str = "default",
         clean_ex_first: bool = True,
         date_features: Union[bool, List[str]] = False,
         date_features_to_one_hot: Union[bool, List[str]] = True,
@@ -1276,6 +1289,8 @@ def cross_validation(
         finetune_steps : int (default=0)
             Number of steps used to finetune TimeGPT in the
             new data.
+        finetune_loss : str (default='default')
+            Loss function to use for finetuning. Options are: `default`, `mae`, `mse`, `rmse`, `mape`, and `smape`.
         clean_ex_first : bool (default=True)
             Clean exogenous signal before making forecasts
             using TimeGPT.
@@ -1313,6 +1328,7 @@ def cross_validation(
                 target_col=target_col,
                 level=level,
                 finetune_steps=finetune_steps,
+                finetune_loss=finetune_loss,
                 clean_ex_first=clean_ex_first,
                 validate_token=validate_token,
                 date_features=date_features,
@@ -1333,6 +1349,7 @@ def cross_validation(
                 target_col=target_col,
                 level=level,
                 finetune_steps=finetune_steps,
+                finetune_loss=finetune_loss,
                 clean_ex_first=clean_ex_first,
                 validate_token=validate_token,
                 date_features=date_features,