Support LLaDa-Image

docs/source/en/_toctree.yml

@@ -379,6 +379,8 @@
         title: Krea2Transformer2DModel
       - local: api/models/latte_transformer3d
         title: LatteTransformer3DModel
+      - local: api/models/llada_image_transformer2d
+        title: LLaDA-Image
       - local: api/models/longcat_image_transformer2d
         title: LongCatImageTransformer2DModel
       - local: api/models/ltx2_video_transformer3d
@@ -617,6 +619,8 @@
         title: Latent Diffusion
       - local: api/pipelines/ledits_pp
         title: LEDITS++
+      - local: api/pipelines/llada_image
+        title: LLaDA-Image
       - local: api/pipelines/longcat_image
         title: LongCat-Image
       - local: api/pipelines/lumina2

docs/source/en/api/models/llada_image_transformer2d.md

@@ -0,0 +1,23 @@
+# LLaDA-Image
+
+The LLaDA-Image model family combines a denoising transformer with a QueryFormer, text projection model, and SigVQ
+image tokenizer. Together, these components support text-to-image generation, VQ-conditioned generation, and
+instruction-guided image editing.
+
+The original code and checkpoints are available in the [LLaDA-Image repository](https://github.com/inclusionAI/LLaDA-Image).
+
+## LLaDAImageTransformer2DModel
+
+[[autodoc]] LLaDAImageTransformer2DModel
+
+## LLaDAImageQueryFormerModel
+
+[[autodoc]] LLaDAImageQueryFormerModel
+
+## LLaDAImageTextProjectionModel
+
+[[autodoc]] LLaDAImageTextProjectionModel
+
+## LLaDAImageSigVQModel
+
+[[autodoc]] LLaDAImageSigVQModel

docs/source/en/api/pipelines/llada_image.md

@@ -0,0 +1,57 @@
+# LLaDA-Image
+
+[LLaDA-Image](https://huggingface.co/inclusionAI/LLaDA-Image) is a unified image generation and editing model. The
+same pipeline supports text-to-image generation, VQ-conditioned generation, and instruction-guided editing with a
+reference image. The Base checkpoint is designed for 50 sampling steps, while
+[LLaDA-Image-Turbo](https://huggingface.co/inclusionAI/LLaDA-Image-Turbo) is distilled for 4 steps.
+
+The checkpoint includes a custom LLaDA2 text encoder, so pass `trust_remote_code=True` when loading it.
+
+```python
+import torch
+
+from diffusers import LLaDAImagePipeline
+
+pipe = LLaDAImagePipeline.from_pretrained(
+    "inclusionAI/LLaDA-Image",
+    dtype=torch.bfloat16,
+    trust_remote_code=True,
+)
+pipe.enable_model_cpu_offload()
+
+image = pipe(
+    prompt="A cinematic photograph of a red fox standing in fresh snow",
+    height=1024,
+    width=1024,
+    num_inference_steps=50,
+    guidance_scale=5.0,
+    generator=torch.Generator("cuda").manual_seed(42),
+).images[0]
+```
+
+For image editing, pass a reference image and select the editing mode.
+
+```python
+from diffusers.utils import load_image
+
+reference_image = load_image("https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/diffusers/cat.png")
+image = pipe(
+    prompt="Turn it into a watercolor painting",
+    image=reference_image,
+    generation_mode="editing",
+    height=1024,
+    width=1024,
+    num_inference_steps=50,
+    guidance_scale=5.0,
+).images[0]
+```
+
+## LLaDAImagePipeline
+
+[[autodoc]] LLaDAImagePipeline
+  - all
+  - call
+
+## LLaDAImagePipelineOutput
+
+[[autodoc]] pipelines.LLaDAImagePipelineOutput

src/diffusers/__init__.py

@@ -309,6 +309,10 @@
             "Kandinsky5Transformer3DModel",
             "Krea2Transformer2DModel",
             "LatteTransformer3DModel",
+            "LLaDAImageQueryFormerModel",
+            "LLaDAImageSigVQModel",
+            "LLaDAImageTextProjectionModel",
+            "LLaDAImageTransformer2DModel",
             "LongCatAudioDiTTransformer",
             "LongCatAudioDiTVae",
             "LongCatImageTransformer2DModel",
@@ -727,6 +731,8 @@
             "LEditsPPPipelineStableDiffusionXL",
             "LLaDA2Pipeline",
             "LLaDA2PipelineOutput",
+            "LLaDAImagePipeline",
+            "LLaDAImagePipelineOutput",
             "LongCatAudioDiTPipeline",
             "LongCatImageEditPipeline",
             "LongCatImagePipeline",
@@ -1191,6 +1197,10 @@
             Kandinsky5Transformer3DModel,
             Krea2Transformer2DModel,
             LatteTransformer3DModel,
+            LLaDAImageQueryFormerModel,
+            LLaDAImageSigVQModel,
+            LLaDAImageTextProjectionModel,
+            LLaDAImageTransformer2DModel,
             LongCatAudioDiTTransformer,
             LongCatAudioDiTVae,
             LongCatImageTransformer2DModel,
@@ -1584,6 +1594,8 @@
             LEditsPPPipelineStableDiffusionXL,
             LLaDA2Pipeline,
             LLaDA2PipelineOutput,
+            LLaDAImagePipeline,
+            LLaDAImagePipelineOutput,
             LongCatAudioDiTPipeline,
             LongCatImageEditPipeline,
             LongCatImagePipeline,

src/diffusers/models/__init__.py

@@ -131,6 +131,12 @@
     _import_structure["transformers.transformer_joyimage_edit_plus"] = ["JoyImageEditPlusTransformer3DModel"]
     _import_structure["transformers.transformer_kandinsky"] = ["Kandinsky5Transformer3DModel"]
     _import_structure["transformers.transformer_krea2"] = ["Krea2Transformer2DModel"]
+    _import_structure["transformers.transformer_llada_image"] = [
+        "LLaDAImageQueryFormerModel",
+        "LLaDAImageSigVQModel",
+        "LLaDAImageTextProjectionModel",
+        "LLaDAImageTransformer2DModel",
+    ]
     _import_structure["transformers.transformer_longcat_audio_dit"] = ["LongCatAudioDiTTransformer"]
     _import_structure["transformers.transformer_longcat_image"] = ["LongCatImageTransformer2DModel"]
     _import_structure["transformers.transformer_ltx"] = ["LTXVideoTransformer3DModel"]
@@ -273,6 +279,10 @@
             Kandinsky5Transformer3DModel,
             Krea2Transformer2DModel,
             LatteTransformer3DModel,
+            LLaDAImageQueryFormerModel,
+            LLaDAImageSigVQModel,
+            LLaDAImageTextProjectionModel,
+            LLaDAImageTransformer2DModel,
             LongCatAudioDiTTransformer,
             LongCatImageTransformer2DModel,
             LTX2VideoTransformer3DModel,

src/diffusers/models/transformers/__init__.py

@@ -46,6 +46,12 @@
     from .transformer_joyimage_edit_plus import JoyImageEditPlusTransformer3DModel
     from .transformer_kandinsky import Kandinsky5Transformer3DModel
     from .transformer_krea2 import Krea2Transformer2DModel
+    from .transformer_llada_image import (
+        LLaDAImageQueryFormerModel,
+        LLaDAImageSigVQModel,
+        LLaDAImageTextProjectionModel,
+        LLaDAImageTransformer2DModel,
+    )
     from .transformer_longcat_audio_dit import LongCatAudioDiTTransformer
     from .transformer_longcat_image import LongCatImageTransformer2DModel
     from .transformer_ltx import LTXVideoTransformer3DModel