Skip to main content
IMAGE MODELby Black Forest LabsFLUX.1 [dev]

Flux Dev

Black Forest Labs’ 12-billion parameter open-weight model — guidance-distilled from FLUX.1 Pro. Known for exceptional visual quality, strong prompt adherence, and as the most popular open-weight image model on Hugging Face.

Parameters
12 Billion
Resolution
Up to 1MP
Architecture
Rectified Flow Transformer
Released
August 2024
Flux Dev example — cinematic landscapeFlux Dev example — artistic compositionFlux Dev example — portraitFlux Dev example — detailed sceneFlux Dev example — open-weight generation

Capabilities

Exceptional prompt adherence

Follows complex, multi-element prompts with high accuracy — comparable to DALL-E 3 and Midjourney v6 at launch.

Strong text rendering

Generates clean, legible text within images with improved accuracy over previous open-weight models.

Natural lighting and composition

Produces well-balanced lighting effects, realistic shadows, and visually coherent compositions across diverse subjects.

Fine-tuning foundation

The most popular open-weight base for LoRA fine-tunes — over 36,000 community adapters available on Hugging Face.

Flexible aspect ratios

Supports aspect ratios from 3:7 portrait to 7:3 landscape with resolutions up to 1 megapixel.

Local deployment

Open weights available on Hugging Face for local and on-premise deployment via Diffusers or ComfyUI.

Specifications

FeatureDetails
ArchitectureHybrid multimodal and parallel diffusion transformer (rectified flow)
Parameters12 billion
Training methodGuidance distillation from FLUX.1 [pro]
Max resolution~1 megapixel (default 1024×1024)
Aspect ratios3:7 to 7:3 (all intermediate ratios supported)
Recommended steps20–50
Prompt token limit512 tokens
ReleasedAugust 1, 2024

How to use

1

Open the AI Image Generator

Go to the ImagineArt AI Image Generator.
2

Select the model

From the model dropdown, choose Flux Dev.
3

Write your prompt

Write a detailed prompt describing the subject, style, lighting, and composition. Flux Dev responds well to descriptive, structured prompts.
4

Set your aspect ratio

Choose the aspect ratio that fits your project. Flux Dev supports a wide range from portrait to landscape.
5

Generate

Click Generate and review your output. Iterate on the prompt to refine results.

Prompting tips

  • Describe lighting explicitly — Terms like “golden hour”, “studio lighting”, “soft diffused light” significantly shape the output.
  • Be specific about style — Mention the photographic or artistic style: “cinematic”, “editorial”, “National Geographic-style”, or art movements like “impressionist”, “concept art”.
  • Use composition language“close-up”, “wide shot”, “shallow depth of field”, “bokeh background” guide framing.
  • Keep complex scenes structured — For multi-element compositions, describe the primary subject first, then supporting elements, then environment.

Example prompts

A bioluminescent forest at night, glowing mushrooms illuminating the forest floor, misty atmosphere, macro photography style, ethereal blue-green light.
Editorial portrait of a woman in 1970s fashion, warm film grain texture, golden hour lighting, cinematic composition, Kodak portra 400 aesthetic.

Compare models

ModelParametersMax resolutionBest forLicense
Flux Dev12B~1MPFine-tuning, research, creative workNon-commercial (open weights)
Flux Ultra 1.112B4MPPrint-res commercial outputProprietary API
ImagineArt 1.5 Pro4K nativeComposition-heavy professional workCommercial
Seedream 4.012B4K nativeMulti-reference campaignsCommercial
Flux Dev’s open weights are hosted on Hugging Face under a non-commercial license. For commercial projects on ImagineArt, all usage is covered under ImagineArt’s platform licensing.