diff --git a/crawl4ai/model_loader.py b/crawl4ai/model_loader.py
index aa80f673d..68f115d2d 100644
--- a/crawl4ai/model_loader.py
+++ b/crawl4ai/model_loader.py
@@ -75,8 +75,8 @@ def get_home_folder():
 def load_bert_base_uncased():
     from transformers import BertTokenizer, BertModel
 
-    tokenizer = BertTokenizer.from_pretrained("bert-base-uncased", resume_download=None)
-    model = BertModel.from_pretrained("bert-base-uncased", resume_download=None)
+    tokenizer = BertTokenizer.from_pretrained("bert-base-uncased")
+    model = BertModel.from_pretrained("bert-base-uncased")
     model.eval()
     model, device = set_model_device(model)
     return tokenizer, model
@@ -94,8 +94,8 @@ def load_HF_embedding_model(model_name="BAAI/bge-small-en-v1.5") -> tuple:
     """
     from transformers import AutoTokenizer, AutoModel
 
-    tokenizer = AutoTokenizer.from_pretrained(model_name, resume_download=None)
-    model = AutoModel.from_pretrained(model_name, resume_download=None)
+    tokenizer = AutoTokenizer.from_pretrained(model_name)
+    model = AutoModel.from_pretrained(model_name)
     model.eval()
     model, device = set_model_device(model)
     return tokenizer, model
@@ -134,10 +134,8 @@ def load_text_multilabel_classifier():
     #     # return load_spacy_model(), torch.device("cpu")
 
     MODEL = "cardiffnlp/tweet-topic-21-multi"
-    tokenizer = AutoTokenizer.from_pretrained(MODEL, resume_download=None)
-    model = AutoModelForSequenceClassification.from_pretrained(
-        MODEL, resume_download=None
-    )
+    tokenizer = AutoTokenizer.from_pretrained(MODEL)
+    model = AutoModelForSequenceClassification.from_pretrained(MODEL)
     model.eval()
     model, device = set_model_device(model)
     class_mapping = model.config.id2label
diff --git a/tests/unit/test_model_loader_transformers.py b/tests/unit/test_model_loader_transformers.py
new file mode 100644
index 000000000..390b4eaa1
--- /dev/null
+++ b/tests/unit/test_model_loader_transformers.py
@@ -0,0 +1,87 @@
+import sys
+from types import SimpleNamespace
+from unittest.mock import Mock
+
+import pytest
+
+from crawl4ai import model_loader
+
+
+class _FakePretrained:
+    from_pretrained = Mock()
+
+
+@pytest.fixture(autouse=True)
+def reset_pretrained_mock():
+    _FakePretrained.from_pretrained.reset_mock()
+
+
+def _fake_model():
+    model = Mock()
+    model.config.id2label = {}
+    return model
+
+
+def test_bert_loader_uses_current_from_pretrained_api(monkeypatch):
+    tokenizer = object()
+    model = _fake_model()
+    _FakePretrained.from_pretrained.side_effect = [tokenizer, model]
+    monkeypatch.setitem(
+        sys.modules,
+        "transformers",
+        SimpleNamespace(BertTokenizer=_FakePretrained, BertModel=_FakePretrained),
+    )
+    monkeypatch.setattr(model_loader, "set_model_device", lambda value: (value, "cpu"))
+
+    loaded_tokenizer, loaded_model = model_loader.load_bert_base_uncased()
+
+    assert (loaded_tokenizer, loaded_model) == (tokenizer, model)
+    assert _FakePretrained.from_pretrained.call_args_list == [
+        (("bert-base-uncased",), {}),
+        (("bert-base-uncased",), {}),
+    ]
+
+
+def test_embedding_loader_uses_current_from_pretrained_api(monkeypatch):
+    tokenizer = object()
+    model = _fake_model()
+    _FakePretrained.from_pretrained.side_effect = [tokenizer, model]
+    monkeypatch.setitem(
+        sys.modules,
+        "transformers",
+        SimpleNamespace(AutoTokenizer=_FakePretrained, AutoModel=_FakePretrained),
+    )
+    monkeypatch.setattr(model_loader, "set_model_device", lambda value: (value, "cpu"))
+
+    loaded_tokenizer, loaded_model = model_loader.load_HF_embedding_model("example/model")
+
+    assert (loaded_tokenizer, loaded_model) == (tokenizer, model)
+    assert _FakePretrained.from_pretrained.call_args_list == [
+        (("example/model",), {}),
+        (("example/model",), {}),
+    ]
+
+
+def test_multilabel_loader_uses_current_from_pretrained_api(monkeypatch):
+    tokenizer = object()
+    model = _fake_model()
+    _FakePretrained.from_pretrained.side_effect = [tokenizer, model]
+    monkeypatch.setitem(
+        sys.modules,
+        "transformers",
+        SimpleNamespace(
+            AutoTokenizer=_FakePretrained,
+            AutoModelForSequenceClassification=_FakePretrained,
+        ),
+    )
+    monkeypatch.setitem(sys.modules, "torch", SimpleNamespace())
+    monkeypatch.setattr(model_loader, "set_model_device", lambda value: (value, "cpu"))
+
+    classifier, device = model_loader.load_text_multilabel_classifier()
+
+    assert callable(classifier)
+    assert device == "cpu"
+    assert _FakePretrained.from_pretrained.call_args_list == [
+        (("cardiffnlp/tweet-topic-21-multi",), {}),
+        (("cardiffnlp/tweet-topic-21-multi",), {}),
+    ]