FLUX
FLUX by Black Forest Labs is a sophisticated AI tool offering creative solutions. Using a combination of cutting-edge AI models and refined aesthetics, it gener...
Last verified:
What is FLUX?
FLUX is a cutting-edge generative AI model family from Black Forest Labs, a frontier AI research lab focused on visual intelligence. The flagship FLUX.2 model delivers production-grade AI image generation and editing with photorealistic 4MP output, multi-reference image control (up to 10 reference images), and unprecedented detail that closes the gap with real photography. The model excels at text-to-image generation, image editing, character consistency across hundreds of assets, product visualization, and production-ready text rendering with complex typography.
Key features include professional-grade controls like exact color matching via hex codes, reliable spatial reasoning with accurate object positioning and realistic physics, JSON-based control system, pose guidance for precise positioning, generative expand/shrink capabilities, and sub-10-second generation speeds. FLUX.2 supports outputs up to 4MP resolution in any aspect ratio, processes 32K text input tokens, and offers four variants: FLUX.2 [max] for top-tier quality, FLUX.2 [pro] for state-of-the-art quality at maximum speed, FLUX.2 [flex] for maximum precision and typography, and FLUX.2 [dev] with open weights for self-deployment.
FLUX is built for artists, designers, marketers, e-commerce businesses, product visualization teams, UI/UX designers, entertainment/media producers, and enterprises needing production-grade image generation. The platform offers API access for scaling workloads, open weights for custom deployment on your own infrastructure, an in-browser playground for experimentation without code, and enterprise solutions for large organizations. Black Forest Labs, founded by Stable Diffusion pioneers, operates from Freiburg and San Francisco with a team of ~70 scientists and engineers.
FLUX pricing
Pricing model: Free
FLUX uses a simple credit-based pay-per-generation model where 1 credit = $0.01 USD. The same pricing applies to both API and Playground. Model costs vary: FLUX.1 Kontext [pro] costs 4 credits ($0.04/image), FLUX.1 Kontext [max] costs 8 credits ($0.08/image), FLUX1.1 [pro] costs 4 credits ($0.04/image), FLUX1.1 [pro] Ultra costs 6 credits ($0.06/image), FLUX.1 Fill [pro] costs 5 credits ($0.05/image), FLUX.2 [klein] starts at 5 credits (~$0.05/image), FLUX.2 [flex] costs 15 credits ($0.15/image), FLUX.2 [pro] costs 20 credits ($0.20/image), and FLUX.2 [max] costs 25 credits ($0.25/image). FLUX.1 Schnell is free and open-source. Finetuning costs range from 200 credits ($2.00) for short training (<150 steps) to 600 credits ($6.00) for long training (>500 steps). Batch requests multiply the base cost by the number of images. Free tier includes FLUX.1 Schnell (open-source) access.
FLUX pros
- Photorealistic 4MP output quality
- Multi-reference control with up to 10 images
- Best-in-class character consistency
- Production-ready text rendering with complex typography
- Exact color matching via hex codes
- Sub-10-second generation speeds
- 32K text input token context window
- Any aspect ratio support
- Reliable spatial reasoning and physics
- JSON-based control system for precision
- Pose guidance for exact positioning
- Generative expand/shrink capabilities
- Open weights available for self-deployment
- Enterprise-grade consistency for production
- Works for both generation and editing tasks
FLUX cons
- Pay-per-use credit pricing model (no unlimited plans)
- Credits expire based on plan type
- API requires asynchronous polling for results
Frequently asked questions about FLUX
What is FLUX and what does it do?
FLUX is a family of state-of-the-art generative AI models from Black Forest Labs for image generation and editing. FLUX.2, the latest model, combines text-to-image generation with multi-reference image editing capabilities, allowing you to reference up to 10 images simultaneously for best-in-class character and style consistency. It produces photorealistic 4MP images with production-ready text, exact color matching, and reliable spatial reasoning.
How do I get started with FLUX?
You can start in three ways: 1) Try the Playground at playground.bfl.ai to test models instantly in your browser without coding, 2) Use the API (api.bfl.ai) for production workloads by creating an account, adding credits, and making API calls, or 3) Download open weights (FLUX.2 [klein], FLUX.1 [dev]) from Hugging Face or GitHub to deploy on your own infrastructure with full control.
What is the pricing for FLUX?
FLUX uses a credit-based pay-per-image model where 1 credit equals $0.01 USD. Prices range from free for FLUX.1 Schnell (open-source) to 25 credits ($0.25) per image for FLUX.2 [max]. Most models cost 4-20 credits per image. The same pricing applies to both API and Playground. You pay only for what you generate, with batch requests multiplying the base cost by the number of images.
What are the different FLUX.2 variants and which should I choose?
FLUX.2 has four variants: FLUX.2 [max] for maximum performance and highest editing consistency with strongest prompt following (25 credits); FLUX.2 [pro] for state-of-the-art quality at maximum speed and affordable price (20 credits); FLUX.2 [flex] specialized for typography and keeping small details with maximum precision (15 credits); and FLUX.2 [klein] optimized for rapid iteration and prototyping with excellent speed-to-quality ratio (5 credits). For new projects, FLUX.2 is recommended over previous FLUX.1 models.
Can I use FLUX for commercial purposes?
Yes, FLUX supports commercial use. API users have commercial license rights for generated images. For open weights, FLUX.2 [klein] is available under Apache 2.0 license for commercial use, while FLUX.1 [dev] uses FLUX Non-Commercial license. Enterprise customers can add FLUX API and open weights licenses to their products and services with SOC 2 and ISO 27001 compliance.
How does multi-reference control work?
FLUX.2 supports referencing up to 10 images simultaneously for the best character consistency available today. You can use multiple reference images to maintain the same character, style, or product across hundreds of generated assets with production-grade consistency. The model understands how to adapt lighting, physics, and positioning to make references integrate naturally into new scenes.
What resolution and aspect ratios does FLUX support?
FLUX.2 outputs up to 4MP (megapixel) resolution with any aspect ratio. The model is reliable from low-res drafts starting at 400px² up to full 4MP production outputs. This flexibility allows you to generate quick low-res drafts for iteration and then create high-resolution final outputs in any dimension you need.
Can I fine-tune FLUX models?
Yes, finetuning is available for FLUX 1.1 [pro] and FLUX.1 [pro] models only (FLUX.1 Kontext [pro] does not support finetuning). Finetuning costs range from 200 credits ($2.00) for short training under 150 steps, to 400 credits ($4.00) for medium training (150-500 steps), to 600 credits ($6.00) for long training over 500 steps. This allows custom training on your datasets for specialized use cases.