stable-diffusion-3.5-large-turbo

ADD-distilled Multimodal Diffusion Transformer (MMDiT) text-to-image model. Uses three fixed, pretrained text encoders with QK-normalization. Designed for high-quality output at low step counts (typically 4 steps, guidance 0.0).

This repository is an unmodified redistribution of stabilityai/stable-diffusion-3.5-large-turbo. Weights, configs, and license are preserved verbatim.

Files

• sd3.5_large_turbo.safetensors — single-file checkpoint.
• transformer/, vae/, scheduler/, text_encoder/, text_encoder_2/, text_encoder_3/, tokenizer*/ — diffusers-format components.
• text_encoders/ — single-file text encoders for ComfyUI workflows.
• model_index.json, SD3.5L_Turbo_example_workflow.json — pipeline and workflow configs.
• LICENSE.md — Stability AI Community License (verbatim).

Inference

import torch
from diffusers import StableDiffusion3Pipeline

pipe = StableDiffusion3Pipeline.from_pretrained(
    "cudabenchmarktest/stable-diffusion-3.5-large-turbo",
    torch_dtype=torch.bfloat16,
).to("cuda")

image = pipe(
    "A capybara holding a sign that reads Hello Fast World",
    num_inference_steps=4,
    guidance_scale=0.0,
).images[0]
image.save("out.png")

License and attribution

Governed by the Stability AI Community License Agreement (see LICENSE.md). Permits research, non-commercial use, and commercial use for organizations or individuals with less than $1M USD in total annual revenue. Above that threshold a separate Stability Enterprise license is required.

• Original release: Stability AI (stabilityai/stable-diffusion-3.5-large-turbo).
• This redistribution: weights and configs unmodified, LICENSE preserved, README replaced, demo image files removed. No additional modifications.