diff --git a/mteb/leaderboard/app.py b/mteb/leaderboard/app.py
index 7c686a06a9..02e9ff500c 100644
--- a/mteb/leaderboard/app.py
+++ b/mteb/leaderboard/app.py
@@ -228,6 +228,7 @@ def filter_models(
 lang_select = gr.Dropdown(
     ISO_TO_LANGUAGE,
     value=sorted(default_results.languages),
+    allow_custom_value=True,
     multiselect=True,
     label="Language",
     info="Select languages to include.",
diff --git a/mteb/model_meta.py b/mteb/model_meta.py
index 8c5e5dc744..aa66778824 100644
--- a/mteb/model_meta.py
+++ b/mteb/model_meta.py
@@ -152,6 +152,9 @@ def is_zero_shot_on(self, tasks: Sequence[AbsTask] | Sequence[str]) -> bool | No
         zero-shot or not on the given tasks.
         Returns None if no training data is specified on the model.
         """
+        # If no tasks were specified, we're obviously zero-shot
+        if not tasks:
+            return True
         if self.training_datasets is None:
             return None
         model_datasets = {ds_name for ds_name, splits in self.training_datasets.items()}