Stable Diffusion Team Launches FLUX.1 Open Source Document Generation Model

Emerging technology company Black Forest Labs, founded by the core team of Stable Diffusion, officially released its innovative FLUX.1 text-to-image model suite today, injecting new vitality and potential into the open-source artificial intelligence field.

This release marks a new stage of popularity and functionality for generative AI technology, signaling an important milestone in industry development.

The company, led by Robin Rombach, Patrick Esser, and Andreas Blattmann, has successfully raised $31 million in seed funding, with prominent investment firm Andreessen Horowitz (a16z) leading the round and attracting heavyweight investors in the industry, including Brendan Iribe, Michael Ovitz, and Garry Tan.

Black Forest Labs focuses on the development of cutting-edge deep learning models for image and video generation and is committed to improving the accessibility and transparency of the technology.

The FLUX.1 model suite is available in three forms: the closed-source FLUX.1 [pro] designed for professional users, which provides services through an API; the open-source FLUX.1 [dev] for non-commercial use; and the accelerated FLUX.1 [schnell] optimized for personal and local development under the Apache 2.0 license. All versions are equipped with a powerful 12 billion parameters and feature an innovative hybrid architecture combining multimodal and parallel diffusion transformer blocks.

The industry has responded enthusiastically to this release, with experts pointing out that FLUX.1 is comparable in output quality to popular closed-source models and even surpasses them in certain aspects.

The launch of FLUX.1 is of great significance to the open-source AI field, especially in the context of recent challenges faced by Stability AI. It provides new possibilities for the future of highly accessible and high-quality image generation models, and is expected to drive innovation in multiple fields such as graphic design and scientific visualization.

At the same time, Black Forest Labs emphasizes responsible development and deployment of AI technology, establishing clear usage guidelines that prohibit the use of technology to generate harmful content such as false information and non-consensual images.

In terms of technological innovation, FLUX.1 introduces the "flow matching" method, optimizing diffusion models and combining rotational position embedding and parallel attention layer techniques, significantly improving performance and hardware efficiency. These innovations demonstrate significant advantages in enhancing visual quality, increasing instruction compliance, and enriching output diversity.

For graphic designers, digital artists, and creative industry professionals, FLUX.1 is undoubtedly a powerful tool that can help them easily create diverse and high-quality images. Meanwhile, the open-source and accelerated versions of FLUX.1 will also inspire more applications and integration innovations in various industries.