SD WebUI Forge - Neo - Anima Release Freeze

^{[ You are on the Neo branch. Click Retro to switch. ]
[ Neo | Retro ]}

"Neo" mainly serves as an continuation for the `latest` version of Forge, which was built on Gradio `4.40.0` before lllyasviel became too busy... Additionally, this fork is focused on optimization and usability, with the main goal of being able to run the latest models without any bloatwares.

- Haoming02

My fork of "Neo" is a feature freeze of commit 7465c0e, the addition of Anima. This branch will remain at the release of 2.13 as a baseline for any branches made within this repo. The main goal of this branch is to have a continually working environment without fear of config files being forcefully deleted by new features. The other branch of this repo is dedicated to my own customisations and preset configuration.

Caution

A lot of the original information has been reworded or removed from this README.
Check upstream for a complete breakdown of features.

Features

Click here for an overview of features as of Feb. 2026

Most base features of the original Automatic1111 Webui should still function

New Features

• Support Anima
• Support Flux.2-Klein
• Support Z-Image
• Support Wan 2.2
• Support Qwen-Image / Qwen-Image-Edit
• Support Flux Kontext
• Support Multi-Image Inputs for Qwen-Image-Edit and Flux-Kontext
• Support Nunchaku (SVDQ) Models
• Support Lumina-Image-2.0
• Support Chroma1-HD
• Support uv package manager
• Support SageAttention, FlashAttention, fp16_accumulation, torch._scaled_mm
• Support loading upscalers in half precision
• Support running tile composition on GPU
• Support .avif, .heif, and .jxl image formats
• Implement Triton Kernel for matmul in torch.int8
• Implement Seed Variance Enhancer
• Implement RescaleCFG
• Implement MaHiRo
• Implement Epsilon Scaling
• Rewrite Preset System
• Update spandrel

Removed Features

• SD2
• SD3
• Forge Spaces
• Hypernetworks
• CLIP Interrogator
• Deepbooru Interrogator
• Textual Inversion Training
• Most built-in Extensions
• Some built-in Scripts
• Some Samplers
• Sampler in RadioGroup
• Unix .sh launch scripts

Optimizations

• [Comfy] Rewrite the Backend (memory_management.py, ModelPatcher, attention.py, etc.)
• No longer git clone any repository on fresh install
• Fix memory leak when switching checkpoints
• Speed up launch time
• Improve timer logs
• Remove unused cmd_args
• Remove unused args_parser
• Remove unused shared_options
• Remove legacy codes
• Fix some typos
• Fix automatic Tiled VAE fallback
• Pad conditioning for SDXL
• Remove redundant upscaler codes
• Improve ForgeCanvas
• Optimize upscaler logics
• Optimize certain operations in Spandrel
• Speed up model loading
• Improve memory management
• Improve color correction
• Update the implementation for MultiDiffusion
• Update the implementation for uni_pc and LCM samplers
• Update the implementation of LoRAs
• Revamp settings
• Check for Extension updates in parallel
• Move embeddings folder into models folder
• ControlNet Rewrite
• Disable Refiner by default
• No longer install bitsandbytes by default
• Lint & Format
• Update Pillow
• Update protobuf
• Update to latest PyTorch
• No longer install open-clip twice
• Update some packages to newer versions
• Update recommended Python to 3.13.12
• Many more... ™️

Commandline

Click here for commandline arguments

These flags can be added after the set COMMANDLINE_ARGS= line in the webui-user.bat (separate each flag with space) Use python launch.py --help to see all available flags

• --xformers: Install the xformers package to speed up generation

• --port: Specify a server port to use. Defaults to 7860

• --api: Enable API access

• Add --cuda-malloc, --cuda-stream, --pin-shared-memory to slightly improve the model loading; in certain situations, they may cause OutOfMemory errors instead...

• --uv: Replace the python -m pip calls with uv pip to massively speed up package installation. Requires uv to be installed

• --uv-symlink: Same as above; but additionally pass --link-mode symlink to the commands. Significantly reduces installation size (~7 GB to ~100 MB). Using symlink means it will directly access the packages from the cache folders; refrain from clearing the cache when setting this option

• --model-ref: Points to a central models folder that contains all your models. Said folder should contain subfolders like Stable-diffusion, Lora, VAE, ESRGAN, etc. This simply replaces the models folder, rather than adding on top of it

• --forge-ref-a1111-home: Point to an Automatic1111 installation to load its models folders i.e. Stable-diffusion, text_encoder, etc.

• --forge-ref-comfy-home: Point to a ComfyUI installation to load its models folders i.e. diffusion_models, clip, etc.

• --forge-ref-comfy-yaml: Point to the ComfyUI extra_model_paths.yaml to load its configurations i.e. base_path, checkpoints, etc.

• --sage: Install the sageattention package to speed up generation, this will also attempt to install triton automatically

• --flash: Install the flash_attn package to speed up generation

• --nunchaku: Install the nunchaku package to inference SVDQ models

• --bnb: Install the bitsandbytes package to do low-bits (nf4) inference

• --onnxruntime-gpu: Install the onnxruntime with the latest GPU support

• --fast-fp8: Use the torch._scaled_mm function when the model type is float8_e4m3fn

• --fast-fp16: Enable the allow_fp16_accumulation option

• --autotune: Enable the torch.backends.cudnn.benchmark option, although this can be slow

Installation

Instructions for those who need them

1. Install git
2. Clone the Repo
3. Setup the environment using only one of these two methods:

3a. Using UV

• Install uv
• Set up venv

  cd sd-webui-forge-neo
  uv venv venv --python 3.13 --seed

• Add the --uv flag to webui-user.bat

3b. Install Python [deprecated method]

• Get Python 3.13.12
  • Remember to enable Add Python to PATH

4. (Optional) Configure Commandline
5. (Optional) Check out Extra Installations in Haoming02's wiki for how to install git, uv, and FFmpeg.
6. Launch the WebUI via webui-user.bat
7. During the first launch, it will automatically install all the requirements
8. Once the installation is finished, the WebUI will start in a browser automatically

Attention Functions

Caution

Nowadays the native PyTorch scaled_dot_product_attention is usually as fast, and also more stable.

Commandline options

Do not just blindly install all of these. The --xformers, --flash, and --sage args are only responsible for installing the packages, not whether its respective attention is used (this also means you can remove them once the packages are successfully installed). To skip a specific attention, add the respective disable arg such as --disable-sage. Forge Neo tries to import the packages and automatically choose the first available attention function in the following order:

1. SageAttention
2. FlashAttention
3. xformers
4. PyTorch
5. Basic

Regarding Issues

Tip

Please try and install a fresh installation from Haoming02's original repo if you encounter errors.
Check out the Wiki that Haoming02 provides as it may contain the solution to your problem.

---

Special thanks to: AUTOMATIC1111, lllyasviel, comfyanonymous, kijai, city96 and Haoming02
along with the rest of the contributors
for their invaluable efforts in the open-source image generation community.