Skip to main content

​
About the Provider

Black Forest Labs is an AI research company founded by the creators of the original Stable Diffusion model. They focus on building state-of-the-art open-weight image generation models, with FLUX.1 as their flagship model family combining frontier image quality with open-weight accessibility for research and commercial use.

​
Model Quickstart

This section helps you quickly get started with the flux-dev model on the Qubrid AI inferencing platform. To use this model, you need:
  • A valid Qubrid API key
  • Access to the Qubrid inference API
  • Basic knowledge of making API requests in your preferred language
Once authenticated with your API key, you can send inference requests to the flux-dev model and receive responses based on your input prompts. Below are example placeholders showing how the model can be accessed using different programming environments.
You can choose the one that best fits your workflow. 
import requests
import json

url = "https://platform.qubrid.com/v1/images/generations"
headers = {
    "Authorization": "Bearer QUBRID_API_KEY",
    "Content-Type": "application/json"
}

data = {
  "model": "flux-dev",
  "prompt": "cinematic shot of a lone astronaut standing on a desolate alien planet, glowing orange sunset sky, dust storms swirling, dramatic lighting, ultra-wide lens composition, movie still aesthetic, realistic space suit details, volumetric atmosphere, 8k sci-fi film scene",
  "num_inference_steps": 28,
  "guidance": 3.5,
  "seed": -1,
  "aspect_ratio": "1:1",
  "image_size": 1024,
  "output_format": "jpg",
  "output_quality": 80
}

response = requests.post(url, headers=headers, json=data)

if response.status_code == 200:
    with open("generated_image.png", "wb") as f:
        f.write(response.content)
        print("Image saved to generated_image.png")
else:
    print(f"Error: {response.status_code}")
    print(response.text)

​
Model Overview

FLUX.1 [dev] is a 12 billion parameter rectified flow transformer developed by Black Forest Labs.
  • It uses a hybrid architecture combining MMDiT (Multi-Modal Diffusion Transformer) and SingleDiT blocks, with dual text encoders — CLIP ViT-L/14 (77 tokens) for global semantic alignment and T5-v1.1-XXL (up to 512 tokens) for rich, nuanced language understanding.
  • A 16-channel VAE (4× more channels than SDXL) enables higher fidelity latent representations, and the model uses Rotary Positional Encoding (RoPE) with a Flow Matching Euler Discrete scheduler. It is guidance-distilled from FLUX.1 [pro], achieving near-pro quality at significantly lower inference cost.

​
Model at a Glance

FeatureDetails
Model IDflux-dev
ProviderBlack Forest Labs
ArchitectureRectified flow transformer — hybrid MMDiT + SingleDiT blocks with RoPE and Flow Matching Euler Discrete scheduler
Model Size12B parameters (~23GB in bf16)
Context LengthN/A
Release DateAugust 2024
LicenseFLUX.1-dev Non-Commercial License
Training DataNot publicly disclosed

​
When to use?

You should consider using FLUX.1 [dev] if:
  • You need photorealistic image generation
  • Your application requires digital illustration and concept art
  • You are creating marketing and branding creatives
  • Your use case involves photography and portrait generation
  • You need a base model for research and fine-tuning with LoRA or ControlNet

​
Inference Parameters

Parameter NameTypeDefaultDescription
Inference Stepsnumber28Number of denoising steps. More steps yield higher quality but slower generation.
Guidance Scalenumber3.5How closely the model follows the prompt. Higher values produce more literal interpretation of the text.
Seednumber-1Random seed for reproducible generation. Use -1 for random.
Aspect Ratiostring1:1Aspect ratio of the output image. Options: 1:1, 16:9, 21:9, 3:2, 2:3, 4:5, 5:4, 3:4, 4:3, 9:16, 9:21.
Image Sizenumber1024Base size in pixels for the longest side of the output image.
Output FormatstringjpgFormat of the generated image. Options: png, jpg, webp.
Output Qualitynumber80Compression quality for jpg/webp output (1–100). Higher values retain more detail.

​
Key Features

  • 12B Parameters with State-of-the-Art Quality: Dual text encoders (CLIP L/14 + T5-v1.1-XXL) for deep prompt understanding and high-fidelity output.
  • 16-Channel VAE: 4× more channels than SDXL for higher fidelity latent image representations.
  • Hybrid MMDiT + SingleDiT Architecture: Combines multi-modal and single-stream diffusion transformer blocks for efficient high-quality generation.
  • Wide Aspect Ratio and Resolution Support: Supports 11 aspect ratios and flexible image sizes up to 1024px.
  • Open Weights: Compatible with LoRA, ControlNet, and fine-tuning for domain-specific customization.
  • Guidance Distilled from FLUX.1 [pro]: Near-pro quality at significantly lower inference cost.

​
Summary

FLUX.1 [dev] is Black Forest Labs’ open-weight rectified flow transformer built for photorealistic image generation.
  • It uses a hybrid MMDiT + SingleDiT architecture with 12B parameters, dual text encoders, and a 16-channel VAE for high-fidelity outputs.
  • It supports 11 aspect ratios, flexible image sizes, and is compatible with LoRA and ControlNet for fine-tuning.
  • The model is guidance-distilled from FLUX.1 [pro], delivering near-pro quality at lower inference cost.
  • Licensed under the FLUX.1-dev Non-Commercial License.