FLUX: The New AI Text-to-image Generator from Black Forest Labs
On August 1st, AI startup Black Forest Labs announced the launch of the company and the release of their first text-to-image AI model package, FLUX.1. The Germany-based company was founded by researchers who developed Stable Diffusion technology and invented the Latent Diffusion technique, aiming to create advanced generative AI for images and videos.
The launch of FLUX.1 occurred approximately seven weeks after the troubled release of Stability AI's Stable Diffusion 3 Medium in mid-June. Stability AI's offering was widely criticized among image synthesis enthusiasts for its poor performance in generating human anatomy, with users sharing examples of distorted limbs and bodies on social media. This problematic launch followed the departure of three key engineers—Robin Rombach, Andreas Blattmann, and Dominik Lorenz—from Stability AI; these engineers, along with Latent Diffusion co-developer Patrick Esser and others, went on to found Black Forest Labs.
Black Forest Labs launched with three FLUX.1 text-to-image models: a high-quality commercial "pro" version, an open-weight mid-range "dev" version for non-commercial use, and a faster open-weight "schnell" version ("schnell" means fast in German). Black Forest Labs claims that their models surpass existing options like Midjourney and DALL-E in terms of image quality and adherence to text prompts.
FLUX.1 was launched as a package of text-to-image models that set a new standard in text-to-image synthesis for image detail, prompt fidelity, style diversity, and scene complexity.
To balance accessibility and model capabilities, FLUX.1 is offered in three variants: FLUX.1 [pro], FLUX.1 [dev], and FLUX.1 [schnell]:
FLUX.1 [pro]: The best of FLUX.1, offering state-of-the-art performance with top-level prompt fidelity, visual quality, image detail, and output diversity.
FLUX.1 [dev]: An open-weight guided distillation model for non-commercial applications. Directly distilled from FLUX.1 [pro], it offers similar quality and prompt fidelity capabilities, being more efficient than a standard model of the same size.
FLUX.1 [schnell]: The fastest model, designed for local development and personal use.
If you also want to produce unique and high-quality images using FLUX.1 models, we invite you to join PUQ. With PUQ, you can access the exceptional image details, prompt fidelity, style diversity, and scene complexity offered by FLUX.1.
With the integration of PUQ’s advanced inference engine, you can run FLUX models up to twice as fast as with eager torch. This means quicker processing times while preserving exceptional quality and detail.
Key Features:
- Superior Image Quality: Create stunning visuals at higher resolutions.
- Advanced Human Anatomy and Photorealism: Generate highly realistic and anatomically precise images.
- Enhanced Prompt Fidelity: Obtain more accurate and relevant images based on your inputs.
- Remarkable Speed: Enjoy the speed and efficiency of Flux Schnell, perfect for high-demand applications.
Transform your imagination into reality using the three different variants of FLUX.1. You can easily access and start using them through the PUQ platform. We are here to guide you on your journey to creating unique visuals. Join now to explore your imagination without limits and produce extraordinary images!
