diff --git a/ultralytics/models/fastsam/prompt.py b/ultralytics/models/fastsam/prompt.py
index 9738252e..89912132 100644
--- a/ultralytics/models/fastsam/prompt.py
+++ b/ultralytics/models/fastsam/prompt.py
@@ -335,12 +335,12 @@ class FastSAMPrompt:
             self.results[0].masks.data = torch.tensor(np.array([onemask]))
         return self.results
 
-    def text_prompt(self, text):
+    def text_prompt(self, text, clip_download_root=None):
         """Processes a text prompt, applies it to existing results and returns the updated results."""
         if self.results[0].masks is not None:
             format_results = self._format_results(self.results[0], 0)
             cropped_images, filter_id, annotations = self._crop_image(format_results)
-            clip_model, preprocess = self.clip.load("ViT-B/32", device=self.device)
+            clip_model, preprocess = self.clip.load("ViT-B/32", download_root=clip_download_root, device=self.device)
             scores = self.retrieve(clip_model, preprocess, cropped_images, text, device=self.device)
             max_idx = torch.argmax(scores)
             max_idx += sum(np.array(filter_id) <= int(max_idx))