FLUX 1.1 - BlackForestLabs

FLUX 1.1 [dev] is a 12-billion-parameter, rectified flow transformer designed to generate images from text prompts. More details are available in our blog post below.

Key Features

High-quality output, just behind the performance of our flagship FLUX.1 [pro] model.
Strong prompt adherence, comparable to leading closed-source alternatives.
Enhanced efficiency from guidance distillation training.
Open weights, enabling new scientific research and inspiring artists to create unique workflows.
Generated images are permitted for personal, scientific, and commercial use, as outlined in the FLUX.1 [dev] Non-Commercial License.

FLUX 1.1 Model Usage

A reference implementation of FLUX.1 [dev], along with sampling code, is available on our dedicated GitHub repository. Developers and creators are encouraged to build from this starting point.

API Access

FLUX.1 models are also accessible via API through:

bfl.ml (currently FLUX.1 [pro])
replicate.com
fal.ai
mystic.ai

ComfyUI

FLUX.1 [dev] is integrated with ComfyUI, offering a node-based workflow for local inference.

Diffusers Library Integration

To use FLUX.1 [dev] with the 🧨 diffusers library in Python, begin by installing or updating diffusers:

pip install -U diffusers

You can then use FluxPipeline to run the model:

import torch
from diffusers import FluxPipeline

pipe = FluxPipeline.from_pretrained("black-forest-labs/FLUX.1-dev", torch_dtype=torch.bfloat16)
pipe.enable_model_cpu_offload()  # Offload to CPU to save VRAM; remove if sufficient GPU power is available

prompt = "A cat holding a sign that says hello world"
image = pipe(
    prompt,
    height=1024,
    width=1024,
    guidance_scale=3.5,
    num_inference_steps=50,
    max_sequence_length=512,
    generator=torch.Generator("cpu").manual_seed(0)
).images[0]
image.save("flux-dev.png")

For additional guidance, consult the diffusers documentation.

Limitations

This model is not designed to provide factual information.
As a statistical model, it may reflect existing societal biases.
Output may occasionally deviate from the given prompt.
Results are sensitive to prompt phrasing.

Prohibited Uses

The model and its derivatives must not be used:

In violation of any applicable national, federal, state, local, or international laws or regulations.
To exploit, harm, or attempt to exploit or harm minors in any way, including creating, distributing, or soliciting child exploitative content.
For generating or spreading false information intended to harm.
To produce or share personally identifiable information with harmful intent.
For harassment, abuse, threats, stalking, or bullying.
To create non-consensual or illegal pornographic content.
For fully automated decision-making that negatively impacts an individual’s legal rights or creates enforceable obligations.
To facilitate or conduct large-scale disinformation campaigns.

License

This model is covered by the FLUX.1 [dev] Non-Commercial License.

Explore other models in our Blog.