DN6 · Aug 19, 2026 · Jul 29, 2026 · Jul 29, 2026 · Jul 29, 2026 · Jul 30, 2026
diff --git a/.github/workflows/pr_tests.yml b/.github/workflows/pr_tests.yml
@@ -114,6 +114,9 @@ jobs:
       run:
         shell: bash
 
+    env:
+      HF_TOKEN: ${{ secrets.DIFFUSERS_HF_HUB_READ_TOKEN }}
+
     steps:
     - name: Checkout diffusers
       uses: actions/checkout@v6
@@ -242,6 +245,9 @@ jobs:
       run:
         shell: bash
 
+    env:
+      HF_TOKEN: ${{ secrets.DIFFUSERS_HF_HUB_READ_TOKEN }}
+
     steps:
     - name: Checkout diffusers
       uses: actions/checkout@v6

diff --git a/src/diffusers/loaders/single_file_utils.py b/src/diffusers/loaders/single_file_utils.py
@@ -149,6 +149,10 @@
         "net.pos_embedder.dim_spatial_range",
     ],
     "flux2": ["model.diffusion_model.single_stream_modulation.lin.weight", "single_stream_modulation.lin.weight"],
+    "chroma": [
+        "model.diffusion_model.distilled_guidance_layer.in_proj.bias",
+        "distilled_guidance_layer.in_proj.bias",
+    ],
     "ltx2": [
         "model.diffusion_model.av_ca_a2v_gate_adaln_single.emb.timestep_embedder.linear_1.weight",
         "vae.per_channel_statistics.mean-of-means",
@@ -204,6 +208,7 @@
     "flux-depth": {"pretrained_model_name_or_path": "black-forest-labs/FLUX.1-Depth-dev"},
     "flux-schnell": {"pretrained_model_name_or_path": "black-forest-labs/FLUX.1-schnell"},
     "flux-2-dev": {"pretrained_model_name_or_path": "black-forest-labs/FLUX.2-dev"},
+    "chroma": {"pretrained_model_name_or_path": "lodestones/Chroma1-HD"},
     "ltx-video": {"pretrained_model_name_or_path": "diffusers/LTX-Video-0.9.0"},
     "ltx-video-0.9.1": {"pretrained_model_name_or_path": "diffusers/LTX-Video-0.9.1"},
     "ltx-video-0.9.5": {"pretrained_model_name_or_path": "Lightricks/LTX-Video-0.9.5"},
@@ -683,6 +688,9 @@ def infer_diffusers_model_type(checkpoint):
     elif any(key in checkpoint for key in CHECKPOINT_KEY_NAMES["flux2"]):
         model_type = "flux-2-dev"
 
+    elif any(key in checkpoint for key in CHECKPOINT_KEY_NAMES["chroma"]):
+        model_type = "chroma"
+
     elif any(key in checkpoint for key in CHECKPOINT_KEY_NAMES["flux"]):
         if any(
             g in checkpoint for g in ["guidance_in.in_layer.bias", "model.diffusion_model.guidance_in.in_layer.bias"]
@@ -754,17 +762,19 @@ def infer_diffusers_model_type(checkpoint):
 
     elif any(key in checkpoint for key in CHECKPOINT_KEY_NAMES["wan"]):
         if "model.diffusion_model.patch_embedding.weight" in checkpoint:
-            target_key = "model.diffusion_model.patch_embedding.weight"
+            prefix = "model.diffusion_model."
         else:
-            target_key = "patch_embedding.weight"
+            prefix = ""
+
+        target_key = f"{prefix}patch_embedding.weight"
 
-        if CHECKPOINT_KEY_NAMES["wan_vace"] in checkpoint:
+        if f"{prefix}{CHECKPOINT_KEY_NAMES['wan_vace']}" in checkpoint:
             if checkpoint[target_key].shape[0] == 1536:
                 model_type = "wan-vace-1.3B"
             elif checkpoint[target_key].shape[0] == 5120:
                 model_type = "wan-vace-14B"
 
-        if CHECKPOINT_KEY_NAMES["wan_animate"] in checkpoint:
+        elif f"{prefix}{CHECKPOINT_KEY_NAMES['wan_animate']}" in checkpoint:
             model_type = "wan-animate-14B"
 
         elif checkpoint[target_key].shape[0] == 1536:

diff --git a/tests/models/autoencoders/test_models_autoencoder_dc.py b/tests/models/autoencoders/test_models_autoencoder_dc.py
@@ -20,7 +20,13 @@
 from diffusers.utils.torch_utils import randn_tensor
 
 from ...testing_utils import IS_GITHUB_ACTIONS, enable_full_determinism, torch_device
-from ..testing_utils import BaseModelTesterConfig, MemoryTesterMixin, ModelTesterMixin, TrainingTesterMixin
+from ..testing_utils import (
+    BaseModelTesterConfig,
+    MemoryTesterMixin,
+    ModelTesterMixin,
+    SingleFileTesterMixin,
+    TrainingTesterMixin,
+)
 from .testing_utils import NewAutoencoderTesterMixin
 
 
@@ -102,3 +108,13 @@ def test_layerwise_casting_memory(self):
 
 class TestAutoencoderDCSlicingTiling(AutoencoderDCTesterConfig, NewAutoencoderTesterMixin):
     """Slicing and tiling tests for AutoencoderDC."""
+
+
+class TestAutoencoderDCSingleFile(AutoencoderDCTesterConfig, SingleFileTesterMixin):
+    @property
+    def ckpt_path(self):
+        return "https://huggingface.co/mit-han-lab/dc-ae-f32c32-sana-1.0/blob/main/model.safetensors"
+
+    @property
+    def pretrained_model_name_or_path(self):
+        return "mit-han-lab/dc-ae-f32c32-sana-1.0-diffusers"
diff --git a/tests/models/autoencoders/test_models_autoencoder_kl.py b/tests/models/autoencoders/test_models_autoencoder_kl.py
@@ -35,7 +35,13 @@
     torch_all_close,
     torch_device,
 )
-from ..testing_utils import BaseModelTesterConfig, MemoryTesterMixin, ModelTesterMixin, TrainingTesterMixin
+from ..testing_utils import (
+    BaseModelTesterConfig,
+    MemoryTesterMixin,
+    ModelTesterMixin,
+    SingleFileTesterMixin,
+    TrainingTesterMixin,
+)
 from .testing_utils import NewAutoencoderTesterMixin
 
 
@@ -428,3 +434,19 @@ def test_stable_diffusion_encode_sample(self, seed, expected_slice):
 
         tolerance = 3e-3 if torch_device != "mps" else 1e-2
         assert torch_all_close(output_slice, expected_output_slice, atol=tolerance)
+
+
+class TestAutoencoderKLSingleFile(AutoencoderKLTesterConfig, SingleFileTesterMixin):
+    @property
+    def ckpt_path(self):
+        return "https://huggingface.co/stabilityai/sd-vae-ft-mse-original/blob/main/vae-ft-mse-840000-ema-pruned.safetensors"
+
+    @property
+    def pretrained_model_name_or_path(self):
+        # `from_single_file` resolves a bare VAE checkpoint against the SD 1.5 repo, so compare against that
+        # rather than against `stabilityai/sd-vae-ft-mse`, whose own config differs (sample_size 256 vs 512).
+        return "stable-diffusion-v1-5/stable-diffusion-v1-5"
+
+    @property
+    def pretrained_model_kwargs(self):
+        return {"subfolder": "vae"}
diff --git a/tests/models/autoencoders/test_models_autoencoder_ltx_video.py b/tests/models/autoencoders/test_models_autoencoder_ltx_video.py
@@ -20,7 +20,13 @@
 from diffusers.utils.torch_utils import randn_tensor
 
 from ...testing_utils import enable_full_determinism, torch_device
-from ..testing_utils import BaseModelTesterConfig, MemoryTesterMixin, ModelTesterMixin, TrainingTesterMixin
+from ..testing_utils import (
+    BaseModelTesterConfig,
+    MemoryTesterMixin,
+    ModelTesterMixin,
+    SingleFileTesterMixin,
+    TrainingTesterMixin,
+)
 from .testing_utils import NewAutoencoderTesterMixin
 
 
@@ -179,3 +185,17 @@ def test_gradient_checkpointing_is_applied(self):
 
 class TestAutoencoderKLLTXVideo091Memory(AutoencoderKLLTXVideo091TesterConfig, MemoryTesterMixin):
     """Memory optimization tests for AutoencoderKLLTXVideo (0.9.1 config)."""
+
+
+class TestAutoencoderKLLTXVideoSingleFile(AutoencoderKLLTXVideo090TesterConfig, SingleFileTesterMixin):
+    @property
+    def ckpt_path(self):
+        return "https://huggingface.co/Lightricks/LTX-Video/blob/main/ltx-video-2b-v0.9.safetensors"
+
+    @property
+    def pretrained_model_name_or_path(self):
+        return "diffusers/LTX-Video-0.9.0"
+
+    @property
+    def pretrained_model_kwargs(self):
+        return {"subfolder": "vae"}
diff --git a/tests/models/autoencoders/test_models_autoencoder_wan.py b/tests/models/autoencoders/test_models_autoencoder_wan.py
@@ -20,7 +20,13 @@
 from diffusers.utils.torch_utils import randn_tensor
 
 from ...testing_utils import enable_full_determinism, torch_device
-from ..testing_utils import BaseModelTesterConfig, MemoryTesterMixin, ModelTesterMixin, TrainingTesterMixin
+from ..testing_utils import (
+    BaseModelTesterConfig,
+    MemoryTesterMixin,
+    ModelTesterMixin,
+    SingleFileTesterMixin,
+    TrainingTesterMixin,
+)
 from .testing_utils import NewAutoencoderTesterMixin
 
 
@@ -86,3 +92,17 @@ def test_layerwise_casting_training(self):
 
 class TestAutoencoderKLWanSlicingTiling(AutoencoderKLWanTesterConfig, NewAutoencoderTesterMixin):
     """Slicing and tiling tests for AutoencoderKLWan."""
+
+
+class TestAutoencoderKLWanSingleFile(AutoencoderKLWanTesterConfig, SingleFileTesterMixin):
+    @property
+    def ckpt_path(self):
+        return "https://huggingface.co/Comfy-Org/Wan_2.1_ComfyUI_repackaged/blob/main/split_files/vae/wan_2.1_vae.safetensors"
+
+    @property
+    def pretrained_model_name_or_path(self):
+        return "Wan-AI/Wan2.1-T2V-1.3B-Diffusers"
+
+    @property
+    def pretrained_model_kwargs(self):
+        return {"subfolder": "vae"}