FLUX
Black Forest Labs' image and video models: open weights plus a pay-as-you-go API
Overview
FLUX is the generative model family from Black Forest Labs, the Freiburg lab started by Stable Diffusion researchers. FLUX 3 covers images, video with native audio and robot action prediction. Some models are open weights under Apache 2.0 or the FLUX Kommunity License; the rest are sold through a pay-as-you-go API and self-hosting licences.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 19 Oct 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
- Weights on Hugging Face
- Commercial use under Apache 2.0
- FLUX.2 [klein] 4B is ungated
- Free commercial use for individuals and companies under $5M annual revenue
- Research and non-commercial use
- Larger organisations use a self-hosting licence or the API
- 768px to 4K output
- No subscriptions or seat fees
- FLUX 3 Image Klein from $0.008 an image
- Charged per first and additional megapixel
- [flex] $0.05 per megapixel
- FLUX.1 and Kontext endpoints still listed
- Video to video $0.21–$0.45 a second
- Video Edit [fast] $0.03 a second
- Upscale $0.07–$0.10 per megapixel-second
- Builder: 10K images a month, 1 domain, 10 licensed users
- Platform and Professional: 100K images a month
- Fine-tuning and LoRA rights
Use cases
What you can produce with FLUX
- Text-to-image generation and editing with FLUX 3 Image through the BFL API
- Images generated locally on an 8GB NVIDIA RTX GPU with the open FLUX 3 Image Klein INT8 build
- Video clips of up to 20 seconds with native audio and multilingual dialogue from FLUX 3 Video
- Generated video upscaled to 2K or 4K with FLUX Video Upscale
- Fine-tuned or LoRA-adapted models run on your own infrastructure under a self-hosting licence
- Task endpoints for outpainting, erasing, deblurring and virtual try-on
- Robot control policies from the open-weight FLUX 3 Action model
FLUX is one of the strongest model families to build on, and since July 2026 it is more than an image model: FLUX 3 adds video with native audio and an open robotics model, while FLUX 3 Image Klein is a small open-weight image model that runs on an 8GB consumer GPU. The API is plain pay-as-you-go pricing with no subscription.
Read the licences before shipping: only the Apache 2.0 models are free for any commercial use, the Kommunity License stops at $5 million in annual revenue, and self-hosting licence prices are not published. Anyone who wants a point-and-click studio rather than a model will still find a third-party app easier.
What we found
FLUX is the model family from Black Forest Labs, the company founded by Robin Rombach, Andreas Blattmann and Patrick Esser, researchers behind Latent Diffusion and Stable Diffusion. Its about page puts its labs in Freiburg and San Francisco; earlier versions of this page said Munich, which is only where the company is registered. The first FLUX.1 models arrived in August 2024. In December 2025 the lab raised a $300 million Series B at a $3.25 billion valuation, co-led by Salesforce Ventures and Anjney Midha's AMP, and its announcement named Adobe, Canva, Meta and Microsoft among companies building on its models.
The family has moved a generation since this entry was first written. FLUX 3, opened in early access on 23 July 2026, is one multimodal model trained on images, video and audio together. FLUX 3 Video became generally available on 4 August with clips up to 20 seconds at HD and Full HD with native audio, and a separate upscaler added 2K and 4K on 20 August. FLUX 3 Action, a 7B open-weight model for robot control, followed in September, and FLUX 3 Image Klein, a 5.6B open-weight image model, on 15 October. The lab says Klein beats larger open models in its own blind evaluations; that is a vendor result.
Licensing needs reading. FLUX.2 klein 4B and FLUX.1 schnell are Apache 2.0. The two open FLUX 3 models use a new FLUX Kommunity License, which allows free commercial use only for individuals and companies under $5 million in annual revenue. FLUX.2 dev and klein 9B remain non-commercial without a paid licence.
The hosted API is pay-as-you-go with no subscription: FLUX 3 Image Klein from $0.008 an image, FLUX 3 Image from $0.041, and FLUX 3 Video from $0.06 a second in draft mode to $0.41 for 4K. Self-hosting licences come in Builder, Platform and Professional tiers whose prices the site does not show. The first-party Playground is a test bench, not a consumer app.
Compiled by RECATOOLS from public sources: vendor documentation and pricing, code repositories and published reporting.
Notable facts
- The Black Forest Labs founding team created the original Stable Diffusion architecture at Ludwig Maximilian University Munich, then commercialised it at Stability AI before leaving to found BFL.
- FLUX.1 was released just 10 days after Black Forest Labs was founded, setting a record for fastest model release by a newly formed AI lab.
- The schnell (German for 'fast') variant generates high-quality images in just 1-4 steps, compared to 20-50 steps for earlier diffusion models.
Frequently asked questions
About this listing
This entry was compiled from publicly available data including FLUX's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with FLUX unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to FLUX directly →
Spotted something out of date? Suggest an update →
Related reading
Alternatives to FLUX
More in Image Generation
