diff --git a/src/datasets/download/streaming_download_manager.py b/src/datasets/download/streaming_download_manager.py
index d702b7ac899..42edd42ce85 100644
--- a/src/datasets/download/streaming_download_manager.py
+++ b/src/datasets/download/streaming_download_manager.py
@@ -692,6 +692,16 @@ def __truediv__(self, p: str) -> "xPath":
         return self.joinpath(p)
 
 
+def xgzip_open(filepath_or_buffer, *args, use_auth_token: Optional[Union[str, bool]] = None, **kwargs):
+    import gzip
+
+    if hasattr(filepath_or_buffer, "read"):
+        return gzip.open(filepath_or_buffer, *args, **kwargs)
+    else:
+        filepath_or_buffer = str(filepath_or_buffer)
+        return gzip.open(xopen(filepath_or_buffer, "rb", use_auth_token=use_auth_token), *args, **kwargs)
+
+
 def xpandas_read_csv(filepath_or_buffer, use_auth_token: Optional[Union[str, bool]] = None, **kwargs):
     import pandas as pd
 
diff --git a/src/datasets/streaming.py b/src/datasets/streaming.py
index 26d7982b53b..cd30f40dcc9 100644
--- a/src/datasets/streaming.py
+++ b/src/datasets/streaming.py
@@ -9,6 +9,7 @@
     xet_parse,
     xgetsize,
     xglob,
+    xgzip_open,
     xisdir,
     xisfile,
     xjoin,
@@ -88,6 +89,7 @@ def wrapper(*args, **kwargs):
     patch_submodule(module, "os.path.getsize", wrap_auth(xgetsize)).start()
     patch_submodule(module, "pathlib.Path", xPath).start()
     # file readers
+    patch_submodule(module, "gzip.open", wrap_auth(xgzip_open)).start()
     patch_submodule(module, "pandas.read_csv", wrap_auth(xpandas_read_csv), attrs=["__version__"]).start()
     patch_submodule(module, "pandas.read_excel", xpandas_read_excel, attrs=["__version__"]).start()
     patch_submodule(module, "scipy.io.loadmat", wrap_auth(xsio_loadmat), attrs=["__version__"]).start()