From b9e1b66d968fd51f9aaf3930c343e69a1e95a5bb Mon Sep 17 00:00:00 2001
From: stevhliu <steven.liu@huggingface.co>
Date: Fri, 10 Jul 2026 09:31:36 -0700
Subject: [PATCH 1/4] alias

---
 src/diffusers/loaders/single_file.py                      | 4 +++-
 src/diffusers/loaders/single_file_model.py                | 4 +++-
 src/diffusers/models/adapter.py                           | 2 +-
 src/diffusers/models/auto_model.py                        | 2 +-
 src/diffusers/models/controlnets/multicontrolnet.py       | 2 +-
 src/diffusers/models/controlnets/multicontrolnet_union.py | 2 +-
 src/diffusers/models/modeling_utils.py                    | 4 +++-
 src/diffusers/modular_pipelines/modular_pipeline_utils.py | 3 ++-
 src/diffusers/pipelines/auto_pipeline.py                  | 8 ++++----
 src/diffusers/pipelines/pipeline_utils.py                 | 8 ++++++--
 10 files changed, 25 insertions(+), 14 deletions(-)

diff --git a/src/diffusers/loaders/single_file.py b/src/diffusers/loaders/single_file.py
index e1cc7f5afe1e..881ff9b96a4c 100644
--- a/src/diffusers/loaders/single_file.py
+++ b/src/diffusers/loaders/single_file.py
@@ -281,7 +281,7 @@ def from_single_file(cls, pretrained_model_link_or_path, **kwargs) -> Self:
                     - A link to the `.ckpt` file (for example
                       `"https://huggingface.co/<repo_id>/blob/main/<path_to_file>.ckpt"`) on the Hub.
                     - A path to a *file* containing all pipeline weights.
-            torch_dtype (`str` or `torch.dtype`, *optional*):
+            dtype (`str` or `torch.dtype`, *optional*):
                 Override the default `torch.dtype` and load the model with another dtype.
             force_download (`bool`, *optional*, defaults to `False`):
                 Whether or not to force the (re-)download of the model weights and configuration files, overriding the
@@ -361,6 +361,8 @@ def from_single_file(cls, pretrained_model_link_or_path, **kwargs) -> Self:
         local_files_only = kwargs.pop("local_files_only", False)
         revision = kwargs.pop("revision", None)
         torch_dtype = kwargs.pop("torch_dtype", None)
+        dtype = kwargs.pop("dtype", None)
+        torch_dtype = dtype if dtype is not None else torch_dtype
         disable_mmap = kwargs.pop("disable_mmap", False)
 
         is_legacy_loading = False
diff --git a/src/diffusers/loaders/single_file_model.py b/src/diffusers/loaders/single_file_model.py
index 9c8508914f3d..56770fd9b6c3 100644
--- a/src/diffusers/loaders/single_file_model.py
+++ b/src/diffusers/loaders/single_file_model.py
@@ -277,7 +277,7 @@ def from_single_file(cls, pretrained_model_link_or_path_or_dict: str | None = No
             original_config (`str`, *optional*):
                 Dict or path to a yaml file containing the configuration for the model in its original format.
                     If a dict is provided, it will be used to initialize the model configuration.
-            torch_dtype (`torch.dtype`, *optional*):
+            dtype (`torch.dtype`, *optional*):
                 Override the default `torch.dtype` and load the model with another dtype.
             force_download (`bool`, *optional*, defaults to `False`):
                 Whether or not to force the (re-)download of the model weights and configuration files, overriding the
@@ -351,6 +351,8 @@ def from_single_file(cls, pretrained_model_link_or_path_or_dict: str | None = No
         revision = kwargs.pop("revision", None)
         config_revision = kwargs.pop("config_revision", None)
         torch_dtype = kwargs.pop("torch_dtype", None)
+        dtype = kwargs.pop("dtype", None)
+        torch_dtype = dtype if dtype is not None else torch_dtype
         quantization_config = kwargs.pop("quantization_config", None)
         low_cpu_mem_usage = kwargs.pop("low_cpu_mem_usage", _LOW_CPU_MEM_USAGE_DEFAULT)
         device = kwargs.pop("device", None)
diff --git a/src/diffusers/models/adapter.py b/src/diffusers/models/adapter.py
index d387859ffbd1..1d146e5575c7 100644
--- a/src/diffusers/models/adapter.py
+++ b/src/diffusers/models/adapter.py
@@ -166,7 +166,7 @@ def from_pretrained(cls, pretrained_model_path: str | os.PathLike | None, **kwar
             pretrained_model_path (`os.PathLike`):
                 A path to a *directory* containing model weights saved using
                 [`~diffusers.models.adapter.MultiAdapter.save_pretrained`], e.g., `./my_model_directory/adapter`.
-            torch_dtype (`torch.dtype`, *optional*):
+            dtype (`torch.dtype`, *optional*):
                 Override the default `torch.dtype` and load the model under this dtype.
             output_loading_info(`bool`, *optional*, defaults to `False`):
                 Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages.
diff --git a/src/diffusers/models/auto_model.py b/src/diffusers/models/auto_model.py
index d3010266ef70..83ba3bca3c1a 100644
--- a/src/diffusers/models/auto_model.py
+++ b/src/diffusers/models/auto_model.py
@@ -173,7 +173,7 @@ def from_pretrained(cls, pretrained_model_or_path: str | os.PathLike | None = No
             cache_dir (`str | os.PathLike`, *optional*):
                 Path to a directory where a downloaded pretrained model configuration is cached if the standard cache
                 is not used.
-            torch_dtype (`torch.dtype`, *optional*):
+            dtype (`torch.dtype`, *optional*):
                 Override the default `torch.dtype` and load the model with another dtype.
             force_download (`bool`, *optional*, defaults to `False`):
                 Whether or not to force the (re-)download of the model weights and configuration files, overriding the
diff --git a/src/diffusers/models/controlnets/multicontrolnet.py b/src/diffusers/models/controlnets/multicontrolnet.py
index 995637c965d3..0ba9f9264f7b 100644
--- a/src/diffusers/models/controlnets/multicontrolnet.py
+++ b/src/diffusers/models/controlnets/multicontrolnet.py
@@ -163,7 +163,7 @@ def from_pretrained(cls, pretrained_model_path: str | os.PathLike | None, **kwar
                 A path to a *directory* containing model weights saved using
                 [`~models.controlnets.multicontrolnet.MultiControlNetModel.save_pretrained`], e.g.,
                 `./my_model_directory/controlnet`.
-            torch_dtype (`torch.dtype`, *optional*):
+            dtype (`torch.dtype`, *optional*):
                 Override the default `torch.dtype` and load the model under this dtype.
             output_loading_info(`bool`, *optional*, defaults to `False`):
                 Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages.
diff --git a/src/diffusers/models/controlnets/multicontrolnet_union.py b/src/diffusers/models/controlnets/multicontrolnet_union.py
index 577492074243..92fe1b0019f2 100644
--- a/src/diffusers/models/controlnets/multicontrolnet_union.py
+++ b/src/diffusers/models/controlnets/multicontrolnet_union.py
@@ -180,7 +180,7 @@ def from_pretrained(cls, pretrained_model_path: str | os.PathLike | None, **kwar
                 A path to a *directory* containing model weights saved using
                 [`~models.controlnets.multicontrolnet.MultiControlNetUnionModel.save_pretrained`], e.g.,
                 `./my_model_directory/controlnet`.
-            torch_dtype (`torch.dtype`, *optional*):
+            dtype (`torch.dtype`, *optional*):
                 Override the default `torch.dtype` and load the model under this dtype.
             output_loading_info(`bool`, *optional*, defaults to `False`):
                 Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages.
diff --git a/src/diffusers/models/modeling_utils.py b/src/diffusers/models/modeling_utils.py
index 650ab4c7fcda..d79994de6efb 100644
--- a/src/diffusers/models/modeling_utils.py
+++ b/src/diffusers/models/modeling_utils.py
@@ -892,7 +892,7 @@ def from_pretrained(cls, pretrained_model_name_or_path: str | os.PathLike | None
             cache_dir (`str | os.PathLike`, *optional*):
                 Path to a directory where a downloaded pretrained model configuration is cached if the standard cache
                 is not used.
-            torch_dtype (`torch.dtype`, *optional*):
+            dtype (`torch.dtype`, *optional*):
                 Override the default `torch.dtype` and load the model with another dtype.
             force_download (`bool`, *optional*, defaults to `False`):
                 Whether or not to force the (re-)download of the model weights and configuration files, overriding the
@@ -1018,6 +1018,8 @@ def from_pretrained(cls, pretrained_model_name_or_path: str | os.PathLike | None
         token = kwargs.pop("token", None)
         revision = kwargs.pop("revision", None)
         torch_dtype = kwargs.pop("torch_dtype", None)
+        dtype = kwargs.pop("dtype", None)
+        torch_dtype = dtype if dtype is not None else torch_dtype
         subfolder = kwargs.pop("subfolder", None)
         device_map = kwargs.pop("device_map", None)
         max_memory = kwargs.pop("max_memory", None)
diff --git a/src/diffusers/modular_pipelines/modular_pipeline_utils.py b/src/diffusers/modular_pipelines/modular_pipeline_utils.py
index 656ab253ccc2..7aa61c55ef1f 100644
--- a/src/diffusers/modular_pipelines/modular_pipeline_utils.py
+++ b/src/diffusers/modular_pipelines/modular_pipeline_utils.py
@@ -311,10 +311,11 @@ def load(self, **kwargs) -> Any:
 
         from diffusers import AutoModel
 
-        # `torch_dtype` is not an accepted parameter for tokenizers and processors.
+        # `dtype`/`torch_dtype` is not an accepted parameter for tokenizers and processors.
         # As a result, it gets stored in `init_kwargs`, which are written to the config
         # during save. This causes JSON serialization to fail when saving the component.
         if self.type_hint is not None and not issubclass(self.type_hint, (torch.nn.Module, AutoModel)):
+            kwargs.pop("dtype", None)
             kwargs.pop("torch_dtype", None)
 
         if self.type_hint is None:
diff --git a/src/diffusers/pipelines/auto_pipeline.py b/src/diffusers/pipelines/auto_pipeline.py
index 5fe97a969a3a..ebeff02e6fbd 100644
--- a/src/diffusers/pipelines/auto_pipeline.py
+++ b/src/diffusers/pipelines/auto_pipeline.py
@@ -426,7 +426,7 @@ def from_pretrained(cls, pretrained_model_or_path, **kwargs):
                     - A path to a *directory* (for example `./my_pipeline_directory/`) containing pipeline weights
                       saved using
                     [`~DiffusionPipeline.save_pretrained`].
-            torch_dtype (`torch.dtype`, *optional*):
+            dtype (`torch.dtype`, *optional*):
                 Override the default `torch.dtype` and load the model with another dtype.
             force_download (`bool`, *optional*, defaults to `False`):
                 Whether or not to force the (re-)download of the model weights and configuration files, overriding the
@@ -717,7 +717,7 @@ def from_pretrained(cls, pretrained_model_or_path, **kwargs):
                     - A path to a *directory* (for example `./my_pipeline_directory/`) containing pipeline weights
                       saved using
                     [`~DiffusionPipeline.save_pretrained`].
-            torch_dtype (`str` or `torch.dtype`, *optional*):
+            dtype (`str` or `torch.dtype`, *optional*):
                 Override the default `torch.dtype` and load the model with another dtype.
             force_download (`bool`, *optional*, defaults to `False`):
                 Whether or not to force the (re-)download of the model weights and configuration files, overriding the
@@ -1022,7 +1022,7 @@ def from_pretrained(cls, pretrained_model_or_path, **kwargs):
                     - A path to a *directory* (for example `./my_pipeline_directory/`) containing pipeline weights
                       saved using
                     [`~DiffusionPipeline.save_pretrained`].
-            torch_dtype (`str` or `torch.dtype`, *optional*):
+            dtype (`str` or `torch.dtype`, *optional*):
                 Override the default `torch.dtype` and load the model with another dtype.
             force_download (`bool`, *optional*, defaults to `False`):
                 Whether or not to force the (re-)download of the model weights and configuration files, overriding the
@@ -1312,7 +1312,7 @@ def from_pretrained(cls, pretrained_model_or_path, **kwargs):
                     - A path to a *directory* (for example `./my_pipeline_directory/`) containing pipeline weights
                       saved using
                     [`~DiffusionPipeline.save_pretrained`].
-            torch_dtype (`torch.dtype`, *optional*):
+            dtype (`torch.dtype`, *optional*):
                 Override the default `torch.dtype` and load the model with another dtype.
             force_download (`bool`, *optional*, defaults to `False`):
                 Whether or not to force the (re-)download of the model weights and configuration files, overriding the
diff --git a/src/diffusers/pipelines/pipeline_utils.py b/src/diffusers/pipelines/pipeline_utils.py
index 7b653caa0657..f1bf0efe4881 100644
--- a/src/diffusers/pipelines/pipeline_utils.py
+++ b/src/diffusers/pipelines/pipeline_utils.py
@@ -639,7 +639,7 @@ def from_pretrained(cls, pretrained_model_name_or_path: str | os.PathLike, **kwa
                       saved using
                     [`~DiffusionPipeline.save_pretrained`].
                     - A path to a *directory* (for example `./my_pipeline_directory/`) containing a dduf file
-            torch_dtype (`torch.dtype` or `dict[str, Union[str, torch.dtype]]`, *optional*):
+            dtype (`torch.dtype` or `dict[str, Union[str, torch.dtype]]`, *optional*):
                 Override the default `torch.dtype` and load the model with another dtype. To load submodels with
                 different dtype pass a `dict` (for example `{'transformer': torch.bfloat16, 'vae': torch.float16}`).
                 Set the default dtype for unspecified components with `default` (for example `{'transformer':
@@ -769,6 +769,8 @@ def from_pretrained(cls, pretrained_model_name_or_path: str | os.PathLike, **kwa
         revision = kwargs.pop("revision", None)
         from_flax = kwargs.pop("from_flax", False)
         torch_dtype = kwargs.pop("torch_dtype", None)
+        dtype = kwargs.pop("dtype", None)
+        torch_dtype = dtype if dtype is not None else torch_dtype
         custom_pipeline = kwargs.pop("custom_pipeline", None)
         custom_revision = kwargs.pop("custom_revision", None)
         provider = kwargs.pop("provider", None)
@@ -2121,7 +2123,9 @@ def from_pipe(cls, pipeline, **kwargs):
         """
 
         original_config = dict(pipeline.config)
-        torch_dtype = kwargs.pop("torch_dtype", torch.float32)
+        torch_dtype = kwargs.pop("torch_dtype", None)
+        dtype = kwargs.pop("dtype", None)
+        torch_dtype = dtype if dtype is not None else (torch_dtype if torch_dtype is not None else torch.float32)
         trust_remote_code = kwargs.pop("trust_remote_code", False)
 
         # derive the pipeline class to instantiate

From 8b6fd02ad328f91c4adb273c457e8fe3f019c9f2 Mon Sep 17 00:00:00 2001
From: stevhliu <steven.liu@huggingface.co>
Date: Mon, 13 Jul 2026 10:36:52 -0700
Subject: [PATCH 2/4] add tests

---
 tests/models/testing_utils/common.py     | 15 +++++++++++++++
 tests/pipelines/test_pipelines_common.py | 20 ++++++++++++++++++++
 2 files changed, 35 insertions(+)

diff --git a/tests/models/testing_utils/common.py b/tests/models/testing_utils/common.py
index 570957cdaf74..59c2e240c1a3 100644
--- a/tests/models/testing_utils/common.py
+++ b/tests/models/testing_utils/common.py
@@ -379,6 +379,21 @@ def test_from_save_pretrained_dtype(self, tmp_path, dtype):
             new_model = self.model_class.from_pretrained(tmp_path, low_cpu_mem_usage=False, torch_dtype=dtype)
             assert new_model.dtype == dtype
 
+    @pytest.mark.parametrize("dtype", [torch.float32, torch.float16, torch.bfloat16], ids=["fp32", "fp16", "bf16"])
+    def test_from_pretrained_dtype_alias(self, tmp_path, dtype):
+        # `dtype` is an alias for `torch_dtype` in `from_pretrained`.
+        model = self.model_class(**self.get_init_dict())
+        model.to(torch_device)
+        model.eval()
+
+        if torch_device == "mps" and dtype == torch.bfloat16:
+            pytest.skip(reason=f"{dtype} is not supported on {torch_device}")
+
+        model.to(dtype)
+        model.save_pretrained(tmp_path)
+        new_model = self.model_class.from_pretrained(tmp_path, low_cpu_mem_usage=True, dtype=dtype)
+        assert new_model.dtype == dtype
+
     @torch.no_grad()
     def test_determinism(self, atol=1e-5, rtol=0):
         model = self.model_class(**self.get_init_dict())
diff --git a/tests/pipelines/test_pipelines_common.py b/tests/pipelines/test_pipelines_common.py
index 139adac2b41e..e70e0fc5c3bd 100644
--- a/tests/pipelines/test_pipelines_common.py
+++ b/tests/pipelines/test_pipelines_common.py
@@ -2408,6 +2408,26 @@ def test_torch_dtype_dict(self):
                     f"Component '{name}' has dtype {component.dtype} but expected {expected_dtype}",
                 )
 
+    def test_dtype_alias(self):
+        # `dtype` is an alias for `torch_dtype` in `from_pretrained`.
+        components = self.get_dummy_components()
+        if not components:
+            self.skipTest("No dummy components defined.")
+
+        pipe = self.pipeline_class(**components)
+
+        with tempfile.TemporaryDirectory(ignore_cleanup_errors=True) as tmpdirname:
+            pipe.save_pretrained(tmpdirname, safe_serialization=False)
+            loaded_pipe = self.pipeline_class.from_pretrained(tmpdirname, dtype=torch.float16)
+
+        for name, component in loaded_pipe.components.items():
+            if isinstance(component, torch.nn.Module) and hasattr(component, "dtype"):
+                self.assertEqual(
+                    component.dtype,
+                    torch.float16,
+                    f"Component '{name}' has dtype {component.dtype} but expected {torch.float16}",
+                )
+
     def test_pipeline_with_accelerator_device_map(self, expected_max_difference=1e-4):
         components = self.get_dummy_components()
         # Set text encoders to eval mode to match from_pretrained behavior

From a0aadf2c9ad9422e7361c1dc987105fde32857b0 Mon Sep 17 00:00:00 2001
From: stevhliu <steven.liu@huggingface.co>
Date: Mon, 13 Jul 2026 10:56:22 -0700
Subject: [PATCH 3/4] feedback

---
 tests/models/testing_utils/common.py     | 6 +++---
 tests/pipelines/test_pipelines_common.py | 3 ---
 2 files changed, 3 insertions(+), 6 deletions(-)

diff --git a/tests/models/testing_utils/common.py b/tests/models/testing_utils/common.py
index 59c2e240c1a3..674421669f3d 100644
--- a/tests/models/testing_utils/common.py
+++ b/tests/models/testing_utils/common.py
@@ -382,13 +382,13 @@ def test_from_save_pretrained_dtype(self, tmp_path, dtype):
     @pytest.mark.parametrize("dtype", [torch.float32, torch.float16, torch.bfloat16], ids=["fp32", "fp16", "bf16"])
     def test_from_pretrained_dtype_alias(self, tmp_path, dtype):
         # `dtype` is an alias for `torch_dtype` in `from_pretrained`.
+        if torch_device == "mps" and dtype == torch.bfloat16:
+            pytest.skip(reason=f"{dtype} is not supported on {torch_device}")
+
         model = self.model_class(**self.get_init_dict())
         model.to(torch_device)
         model.eval()
 
-        if torch_device == "mps" and dtype == torch.bfloat16:
-            pytest.skip(reason=f"{dtype} is not supported on {torch_device}")
-
         model.to(dtype)
         model.save_pretrained(tmp_path)
         new_model = self.model_class.from_pretrained(tmp_path, low_cpu_mem_usage=True, dtype=dtype)
diff --git a/tests/pipelines/test_pipelines_common.py b/tests/pipelines/test_pipelines_common.py
index e70e0fc5c3bd..07aca0156cee 100644
--- a/tests/pipelines/test_pipelines_common.py
+++ b/tests/pipelines/test_pipelines_common.py
@@ -2411,9 +2411,6 @@ def test_torch_dtype_dict(self):
     def test_dtype_alias(self):
         # `dtype` is an alias for `torch_dtype` in `from_pretrained`.
         components = self.get_dummy_components()
-        if not components:
-            self.skipTest("No dummy components defined.")
-
         pipe = self.pipeline_class(**components)
 
         with tempfile.TemporaryDirectory(ignore_cleanup_errors=True) as tmpdirname:

From 5913ca38bd7a10b25b73c1424d6ebf44bacf1017 Mon Sep 17 00:00:00 2001
From: stevhliu <steven.liu@huggingface.co>
Date: Mon, 13 Jul 2026 19:49:09 -0700
Subject: [PATCH 4/4] Fix failing test_dtype_alias pipeline tests

Cosmos pipelines keep the safety_checker in float32 and require it to be
passed explicitly on reload, mirroring their test_torch_dtype_dict
overrides. PRX skips dtype-dict/save-load tests because the custom
T5GemmaEncoder is not compatible with transformers v5; do the same for
test_dtype_alias.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
---
 tests/pipelines/cosmos/test_cosmos.py         | 21 ++++++++++++++++++
 .../cosmos/test_cosmos2_5_predict.py          | 21 ++++++++++++++++++
 .../cosmos/test_cosmos2_5_transfer.py         | 22 +++++++++++++++++++
 .../cosmos/test_cosmos2_text2image.py         | 21 ++++++++++++++++++
 .../cosmos/test_cosmos2_video2world.py        | 21 ++++++++++++++++++
 .../cosmos/test_cosmos_video2world.py         | 21 ++++++++++++++++++
 tests/pipelines/prx/test_pipeline_prx.py      |  4 ++++
 7 files changed, 131 insertions(+)

diff --git a/tests/pipelines/cosmos/test_cosmos.py b/tests/pipelines/cosmos/test_cosmos.py
index 3aa92dca1677..f0407b032f3d 100644
--- a/tests/pipelines/cosmos/test_cosmos.py
+++ b/tests/pipelines/cosmos/test_cosmos.py
@@ -349,6 +349,27 @@ def test_torch_dtype_dict(self):
                     f"Component '{name}' has dtype {component.dtype} but expected {expected_dtype}",
                 )
 
+    def test_dtype_alias(self):
+        # `dtype` is an alias for `torch_dtype` in `from_pretrained`.
+        components = self.get_dummy_components()
+        pipe = self.pipeline_class(**components)
+
+        with tempfile.TemporaryDirectory(ignore_cleanup_errors=True) as tmpdirname:
+            pipe.save_pretrained(tmpdirname, safe_serialization=False)
+            loaded_pipe = self.pipeline_class.from_pretrained(
+                tmpdirname, safety_checker=DummyCosmosSafetyChecker(), dtype=torch.float16
+            )
+
+        for name, component in loaded_pipe.components.items():
+            if name == "safety_checker":
+                continue
+            if isinstance(component, torch.nn.Module) and hasattr(component, "dtype"):
+                self.assertEqual(
+                    component.dtype,
+                    torch.float16,
+                    f"Component '{name}' has dtype {component.dtype} but expected {torch.float16}",
+                )
+
     @unittest.skip(
         "The pipeline should not be runnable without a safety checker. The test creates a pipeline without passing in "
         "a safety checker, which makes the pipeline default to the actual Cosmos Guardrail. The Cosmos Guardrail is "
diff --git a/tests/pipelines/cosmos/test_cosmos2_5_predict.py b/tests/pipelines/cosmos/test_cosmos2_5_predict.py
index 54d4edb485fe..6124e7be4b7b 100644
--- a/tests/pipelines/cosmos/test_cosmos2_5_predict.py
+++ b/tests/pipelines/cosmos/test_cosmos2_5_predict.py
@@ -328,6 +328,27 @@ def test_torch_dtype_dict(self):
                     f"Component '{name}' has dtype {component.dtype} but expected {expected_dtype}",
                 )
 
+    def test_dtype_alias(self):
+        # `dtype` is an alias for `torch_dtype` in `from_pretrained`.
+        components = self.get_dummy_components()
+        pipe = self.pipeline_class(**components)
+
+        with tempfile.TemporaryDirectory(ignore_cleanup_errors=True) as tmpdirname:
+            pipe.save_pretrained(tmpdirname, safe_serialization=False)
+            loaded_pipe = self.pipeline_class.from_pretrained(
+                tmpdirname, safety_checker=DummyCosmosSafetyChecker(), dtype=torch.float16
+            )
+
+        for name, component in loaded_pipe.components.items():
+            if name == "safety_checker":
+                continue
+            if isinstance(component, torch.nn.Module) and hasattr(component, "dtype"):
+                self.assertEqual(
+                    component.dtype,
+                    torch.float16,
+                    f"Component '{name}' has dtype {component.dtype} but expected {torch.float16}",
+                )
+
     @unittest.skip(
         "The pipeline should not be runnable without a safety checker. The test creates a pipeline without passing in "
         "a safety checker, which makes the pipeline default to the actual Cosmos Guardrail. The Cosmos Guardrail is "
diff --git a/tests/pipelines/cosmos/test_cosmos2_5_transfer.py b/tests/pipelines/cosmos/test_cosmos2_5_transfer.py
index c098a46a8f8a..c7f2f799216f 100644
--- a/tests/pipelines/cosmos/test_cosmos2_5_transfer.py
+++ b/tests/pipelines/cosmos/test_cosmos2_5_transfer.py
@@ -426,6 +426,28 @@ def test_torch_dtype_dict(self):
                     f"Component '{name}' has dtype {component.dtype} but expected {expected_dtype}",
                 )
 
+    def test_dtype_alias(self):
+        # `dtype` is an alias for `torch_dtype` in `from_pretrained`.
+        components = self.get_dummy_components()
+        pipe = self.pipeline_class(**components)
+
+        with tempfile.TemporaryDirectory(ignore_cleanup_errors=True) as tmpdirname:
+            pipe.save_pretrained(tmpdirname, safe_serialization=False)
+            loaded_pipe = self.pipeline_class.from_pretrained(
+                tmpdirname, safety_checker=DummyCosmosSafetyChecker(), dtype=torch.float16
+            )
+
+        for name, component in loaded_pipe.components.items():
+            # Skip components that are not loaded from disk or have special handling
+            if name == "safety_checker":
+                continue
+            if isinstance(component, torch.nn.Module) and hasattr(component, "dtype"):
+                self.assertEqual(
+                    component.dtype,
+                    torch.float16,
+                    f"Component '{name}' has dtype {component.dtype} but expected {torch.float16}",
+                )
+
     def test_save_load_optional_components(self, expected_max_difference=1e-4):
         self.pipeline_class._optional_components.remove("safety_checker")
         super().test_save_load_optional_components(expected_max_difference=expected_max_difference)
diff --git a/tests/pipelines/cosmos/test_cosmos2_text2image.py b/tests/pipelines/cosmos/test_cosmos2_text2image.py
index 71c61eff0054..2a7b130c5340 100644
--- a/tests/pipelines/cosmos/test_cosmos2_text2image.py
+++ b/tests/pipelines/cosmos/test_cosmos2_text2image.py
@@ -333,6 +333,27 @@ def test_torch_dtype_dict(self):
                     f"Component '{name}' has dtype {component.dtype} but expected {expected_dtype}",
                 )
 
+    def test_dtype_alias(self):
+        # `dtype` is an alias for `torch_dtype` in `from_pretrained`.
+        components = self.get_dummy_components()
+        pipe = self.pipeline_class(**components)
+
+        with tempfile.TemporaryDirectory(ignore_cleanup_errors=True) as tmpdirname:
+            pipe.save_pretrained(tmpdirname, safe_serialization=False)
+            loaded_pipe = self.pipeline_class.from_pretrained(
+                tmpdirname, safety_checker=DummyCosmosSafetyChecker(), dtype=torch.float16
+            )
+
+        for name, component in loaded_pipe.components.items():
+            if name == "safety_checker":
+                continue
+            if isinstance(component, torch.nn.Module) and hasattr(component, "dtype"):
+                self.assertEqual(
+                    component.dtype,
+                    torch.float16,
+                    f"Component '{name}' has dtype {component.dtype} but expected {torch.float16}",
+                )
+
     @unittest.skip(
         "The pipeline should not be runnable without a safety checker. The test creates a pipeline without passing in "
         "a safety checker, which makes the pipeline default to the actual Cosmos Guardrail. The Cosmos Guardrail is "
diff --git a/tests/pipelines/cosmos/test_cosmos2_video2world.py b/tests/pipelines/cosmos/test_cosmos2_video2world.py
index 1b814257a30a..ca62dcc9665c 100644
--- a/tests/pipelines/cosmos/test_cosmos2_video2world.py
+++ b/tests/pipelines/cosmos/test_cosmos2_video2world.py
@@ -347,6 +347,27 @@ def test_torch_dtype_dict(self):
                     f"Component '{name}' has dtype {component.dtype} but expected {expected_dtype}",
                 )
 
+    def test_dtype_alias(self):
+        # `dtype` is an alias for `torch_dtype` in `from_pretrained`.
+        components = self.get_dummy_components()
+        pipe = self.pipeline_class(**components)
+
+        with tempfile.TemporaryDirectory(ignore_cleanup_errors=True) as tmpdirname:
+            pipe.save_pretrained(tmpdirname, safe_serialization=False)
+            loaded_pipe = self.pipeline_class.from_pretrained(
+                tmpdirname, safety_checker=DummyCosmosSafetyChecker(), dtype=torch.float16
+            )
+
+        for name, component in loaded_pipe.components.items():
+            if name == "safety_checker":
+                continue
+            if isinstance(component, torch.nn.Module) and hasattr(component, "dtype"):
+                self.assertEqual(
+                    component.dtype,
+                    torch.float16,
+                    f"Component '{name}' has dtype {component.dtype} but expected {torch.float16}",
+                )
+
     @unittest.skip(
         "The pipeline should not be runnable without a safety checker. The test creates a pipeline without passing in "
         "a safety checker, which makes the pipeline default to the actual Cosmos Guardrail. The Cosmos Guardrail is "
diff --git a/tests/pipelines/cosmos/test_cosmos_video2world.py b/tests/pipelines/cosmos/test_cosmos_video2world.py
index 925c3b448677..dbe072d334c4 100644
--- a/tests/pipelines/cosmos/test_cosmos_video2world.py
+++ b/tests/pipelines/cosmos/test_cosmos_video2world.py
@@ -362,6 +362,27 @@ def test_torch_dtype_dict(self):
                     f"Component '{name}' has dtype {component.dtype} but expected {expected_dtype}",
                 )
 
+    def test_dtype_alias(self):
+        # `dtype` is an alias for `torch_dtype` in `from_pretrained`.
+        components = self.get_dummy_components()
+        pipe = self.pipeline_class(**components)
+
+        with tempfile.TemporaryDirectory(ignore_cleanup_errors=True) as tmpdirname:
+            pipe.save_pretrained(tmpdirname, safe_serialization=False)
+            loaded_pipe = self.pipeline_class.from_pretrained(
+                tmpdirname, safety_checker=DummyCosmosSafetyChecker(), dtype=torch.float16
+            )
+
+        for name, component in loaded_pipe.components.items():
+            if name == "safety_checker":
+                continue
+            if isinstance(component, torch.nn.Module) and hasattr(component, "dtype"):
+                self.assertEqual(
+                    component.dtype,
+                    torch.float16,
+                    f"Component '{name}' has dtype {component.dtype} but expected {torch.float16}",
+                )
+
     @unittest.skip(
         "The pipeline should not be runnable without a safety checker. The test creates a pipeline without passing in "
         "a safety checker, which makes the pipeline default to the actual Cosmos Guardrail. The Cosmos Guardrail is "
diff --git a/tests/pipelines/prx/test_pipeline_prx.py b/tests/pipelines/prx/test_pipeline_prx.py
index 20c2727b3156..5cd46deadbb5 100644
--- a/tests/pipelines/prx/test_pipeline_prx.py
+++ b/tests/pipelines/prx/test_pipeline_prx.py
@@ -280,3 +280,7 @@ def test_save_load_optional_components(self):
     @unittest.skip("Custom T5GemmaEncoder not compatible with transformers v5.")
     def test_torch_dtype_dict(self):
         pass
+
+    @unittest.skip("Custom T5GemmaEncoder not compatible with transformers v5.")
+    def test_dtype_alias(self):
+        pass