FLUX.1 Dev

FLUX.1 Dev is Black Forest Labs' state-of-the-art text-to-image model. With 12 billion parameters and a rectified flow transformer architecture, it produces images rivaling Midjourney and DALL-E 3. It's the most popular open image model with 704K monthly downloads.

Key features

1. 12B parameters - rectified flow transformer architecture
2. 1024×1024 output - high-resolution image generation
3. Guidance distilled - efficient inference from FLUX.1 Pro
4. ComfyUI support - node-based workflow
5. 41K+ adapters - massive LoRA ecosystem
6. 704K monthly downloads - top text-to-image model

Usage

from diffusers import FluxPipeline
import torch
 
pipe = FluxPipeline.from_pretrained(
    "black-forest-labs/FLUX.1-dev",
    torch_dtype=torch.bfloat16
)
pipe.enable_model_cpu_offload()
 
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,
).images[0]
image.save("output.png")

VRAM requirements

Limitations

• Non-commercial license - personal/research use only
• No factual accuracy - not for information generation
• Prompt dependent - quality varies with prompting style

When to use

• Art & design - concept art, illustrations
• Marketing - social media visuals
• Product design - rapid prototyping
• Creative exploration - style experimentation