This page may contain affiliate links. We may earn a commission if you purchase through our links, at no extra cost to you. Learn more.

FLUX

Black Forest Labs' state-of-the-art open model rivaling Midjourney

4.7/5(0 reviews)

What is FLUX?

FLUX is the image generation model family developed by Black Forest Labs, founded by the original creators of Stable Diffusion. Released in August 2024, FLUX.1 immediately established itself as the strongest open-weight model available, producing images that rival Midjourney V6 in quality while remaining accessible for self-hosting.

The FLUX family includes three variants: FLUX.1 pro (closed API only), FLUX.1 dev (open weights for research and non-commercial use), and FLUX.1 schnell (Apache 2.0 licensed for commercial use). This tiered approach allows developers to experiment freely while offering a premium API for production use cases.

FLUX excels particularly at photorealism and natural human portraits, areas where previous open-source models had struggled. Its architecture uses a hybrid transformer approach that handles long prompt descriptions more effectively than traditional diffusion models. It has rapidly become the preferred base model for fine-tuning in the open-source community.

INFOGRAPHIC FLUX at a Glance
2026 Data
4.7
Overall Rating
10
Key Features
Image Generation
Category
Flow Matching
Architecture
1-4 steps
Schnell Steps
Apache 2.0
License (schnell)
State-of-art
Portrait Quality
Pricing Tiers
Schnell (self-hosted) Free
Dev (research) Free
Pro API ~$0.055/image
BFL API Variable
Founded: 2024 | By: Black Forest Labs FREE PLAN

Key Features

  • Three model tiers: pro API, dev research, schnell commercial
  • State-of-the-art photorealism and portrait quality
  • Superior long prompt comprehension
  • Hybrid transformer architecture using Flow Matching
  • Available on Hugging Face, Replicate, and fal.ai
  • LoRA training support via community tools
  • Fast inference with schnell variant in 1-4 steps
  • FLUX.1 dev open weights for research use
  • Improved text rendering over previous open models
  • Growing fine-tune ecosystem on Civitai

Pros & Cons

Pros

  • Image quality matches or exceeds Midjourney on many prompts
  • Open-weight dev and schnell models available for self-hosting
  • Excellent at natural human portraits and photorealism
  • Apache 2.0 schnell model for commercial use at no cost

Cons

  • Pro model only accessible via paid API
  • Dev model restricted to non-commercial research use
  • Fewer fine-tunes and LoRAs compared to SD ecosystem
  • Higher compute requirements than older SD models

Pricing

Model: Freemium

PlanPriceKey Limits
FLUX.1 schnell self-hosted$0Free, Apache 2.0 commercial license
FLUX.1 dev self-hosted$0Free for non-commercial research only
FLUX.1 pro via APIPay-per-useApproximately $0.055 per image on Replicate or fal.ai
BFL APIPay-per-useDirect API from Black Forest Labs, variable pricing
See Full Pricing

FAQ

Is FLUX free to use?
FLUX.1 schnell is free under Apache 2.0 for commercial use when self-hosted. FLUX.1 dev is free for non-commercial research. The pro variant requires API payment.
Does FLUX have an affiliate program?
Black Forest Labs does not currently offer an affiliate program. Third-party platforms hosting FLUX may have their own programs.
FLUX vs Stable Diffusion: which is better?
FLUX produces significantly better quality images out of the box, especially for photorealism. Stable Diffusion has a much larger ecosystem of fine-tunes and tools accumulated over several years.
Where can I try FLUX online without installing anything?
FLUX is available on Hugging Face Spaces, fal.ai, Replicate, and Freepik Mystic. Several platforms offer free credits to try it.
Can I fine-tune FLUX on my own images?
Yes. FLUX.1 dev can be fine-tuned with LoRA using tools like SimpleTuner. The schnell model is also fine-tunable for commercial use under its Apache 2.0 license.