diff --git a/fast_saver.py b/fast_saver.py
index 06756bd..43c1a0d 100644
--- a/fast_saver.py
+++ b/fast_saver.py
@@ -202,7 +202,8 @@ class FastAbsoluteSaver:
             "hidden": {"prompt": "PROMPT", "extra_pnginfo": "EXTRA_PNGINFO"},
         }
 
-    RETURN_TYPES = ()
+    RETURN_TYPES = ("LATENT",)
+    RETURN_NAMES = ("latent",)
     FUNCTION = "save_images_fast"
     OUTPUT_NODE = True
     CATEGORY = "JSON Dynamic/io"
@@ -546,7 +547,7 @@ class FastAbsoluteSaver:
                 avg_score = (sum(scores_list) / len(scores_list)) if scores_list else None
                 self._save_sidecar_png(img, png_path, avg_score, metadata_key, prompt, extra_pnginfo)
                 print(f"xx- FastSaver: Metadata PNG sidecar saved to {png_path}")
-            return {"ui": {"images": []}}
+            return {"ui": {"images": []}, "result": (latent,)}
 
         if max_threads == 0:
             max_threads = os.cpu_count() or 4
@@ -623,7 +624,7 @@ class FastAbsoluteSaver:
             self._save_sidecar_png(img, png_path, first_score, metadata_key, prompt, extra_pnginfo)
             print(f"xx- FastSaver: Metadata PNG sidecar saved to {png_path}")
 
-        return {"ui": {"images": []}}
+        return {"ui": {"images": []}, "result": (latent,)}
 
 
 class JDL_LoadLatentAbsolute:
diff --git a/tests/test_fast_saver_latent.py b/tests/test_fast_saver_latent.py
index a50b923..7b301e1 100644
--- a/tests/test_fast_saver_latent.py
+++ b/tests/test_fast_saver_latent.py
@@ -45,6 +45,16 @@ def test_png_save_writes_matching_latent_sidecar(tmp_path):
     assert loaded["keep"] == {"value": 7}
 
 
+def test_png_save_returns_latent_passthrough(tmp_path):
+    saver = FastAbsoluteSaver()
+    latent = {"samples": torch.ones((1, 1, 2, 2))}
+
+    result = saver.save_images_fast(**_save_args(tmp_path, latent=latent))
+
+    assert result["result"] == (latent,)
+    assert result["result"][0] is latent
+
+
 def test_video_save_writes_latent_sidecar_next_to_video(tmp_path):
     saver = FastAbsoluteSaver()
     latent = {"samples": torch.arange(8, dtype=torch.float32).reshape(2, 1, 2, 2)}