FLUX: The Rise of Next-Generation AI Image Generation Model by Stable Diffusion Developers

A new revolution is unfolding in the world of AI image generation. FLUX, developed by Black Forest Labs - a company founded by the independent team behind Stable Diffusion - has been making waves in the industry since its announcement in August 2024 with its overwhelming performance. Recording benchmark results that surpass both Midjourney and DALL-E 3 in text generation capabilities and human figure accuracy¹, FLUX has rapidly gained support, particularly within the open-source community.

Black Forest Labs successfully raised $31 million in a seed round from prominent VCs including Andreessen Horowitz and General Catalyst². Subsequently, in January 2025, the company announced a strategic partnership with NVIDIA, with FLUX being adopted as a foundation model for the Blackwell microarchitecture³.

Technical Features and Innovation of FLUX

The FLUX series employs 12 billion parameter rectified flow transformer blocks, achieving performance that sets it apart from conventional models. The following technical innovations are particularly noteworthy:

Model Variations and Features

Model Name	Features	License	Use Cases
FLUX.1 Schnell	Fast generation (seconds)	Apache License (Open Source)	Personal and commercial use
FLUX.1 Dev	High quality, development-oriented	Non-commercial license	Research and development
FLUX.1 Pro	Highest quality	Proprietary	Commercial use via API
FLUX.1 Pro Ultra	4 megapixel support	Proprietary	High-resolution commercial production

The latest FLUX.1 Pro Ultra can generate ultra-high resolution images up to 4 megapixels (4 times the standard model) in approximately 10 seconds, achieving 2.5 times faster performance compared to conventional high-resolution generation⁴.

Technical Advantages

FLUX’s greatest strength lies in its text generation capabilities. While Midjourney has only recently begun supporting text generation, FLUX has achieved high-precision text rendering from its initial versions. It can accurately render complex text, signs, and logos in a single output⁵.

In human figure rendering, FLUX has made significant progress, particularly in hand drawing. While not perfect, it can consistently generate more realistic and proportionally accurate body parts compared to previous open-source models⁶.

The Birth of Black Forest Labs and Its Founders

The founding of Black Forest Labs marked a crucial turning point in the AI image generation industry. Established in early 2024 by Robin Rombach, Andreas Blattmann, and Patrick Esser - who left Stability AI - the company subsequently attracted multiple former colleagues to form a strong founding team⁷.

Founder Profiles

Robin Rombach

Studied physics at Heidelberg University before earning his PhD in computer science. Known as one of the early architects of text-to-image generation models, he’s so renowned in the industry that it’s said “anyone in the field knows Robin Rombach from South Germany”⁸.

Patrick Esser

Worked alongside Rombach at Ludwig Maximilian University of Munich under Professor Björn Ommer. Under the constraints of limited resources, he developed the innovative “latent generation model” approach, which became the foundation for Stable Diffusion⁹.

Andreas Blattmann

Also came from the same research group. The trio published their research findings on image generation in 2022, which led to the birth of Stable Diffusion¹⁰.

Deployment and Use Cases in the Japanese Market

The Japanese AI community has been rapidly adopting FLUX. Many Japanese users are beginning to transition, particularly due to its high compatibility with existing interfaces like ComfyUI and Stable Diffusion WebUI Forge.

On technical sharing platforms like Qiita, numerous implementation guides have been published, such as “Running FLUX.1 Image Generation AI with Black Forest Labs Reference Implementation, Diffusers, and ComfyUI”¹¹, enriching the availability of technical information in Japanese.

Development of Japan-Specific Workflows

Japanese creators have been developing unique workflows using FLUX. For instance, workflows specialized in traditional Japanese art styles such as “Traditional Japanese Alien” and “Japanese ink painting” have been created, gaining attention in the global community¹².

However, comfortable use of FLUX requires high specifications. Japanese tutorials indicate that at least 32GB of RAM is necessary, which has become a barrier to widespread adoption¹³.

Why FLUX is Being Chosen

The rapid expansion of FLUX’s support isn’t solely due to technical superiority. The following factors work in combination:

1. Commitment to Open Source

Black Forest Labs continues the open-weight approach that was key to Stable Diffusion’s success. The FLUX.1 Schnell model is released as fully open source, allowing anyone to use it commercially for free¹⁴.

2. Developer-Friendly Design

The design philosophy prioritizes developers through easy-to-use APIs, compatibility with existing tools, and comprehensive documentation. This makes it easy for many developers to integrate FLUX into their own projects.

3. Continuous Innovation

The company continuously adds innovative features, such as FLUX.1 Tools (inpainting, outpainting, depth map control, etc.) in November 2024¹⁵, and FLUX.1 Kontext (allowing prompts with both images and text) in May 2025¹⁶.

4. Strong Partnerships

The company actively pursues collaboration with major industry players, including integration with Mistral AI’s Le Chat chatbot¹⁷ and strategic partnership with NVIDIA.

Future Outlook and Challenges

FLUX’s success demonstrates a new trend in the AI image generation industry. It has proven that open-source models can achieve quality equal to or better than closed-source commercial models.

However, challenges remain. There are numerous issues to address, including high hardware requirements, steep learning curves, and ethical concerns such as copyright and deepfakes.

Nevertheless, with Black Forest Labs’ innovative approach and support from the global community, FLUX is expected to continue leading the forefront of AI image generation. Particularly in the Japanese market, its high affinity with anime and manga culture suggests potential for unique development.

FLUX represents more than just another AI image generation model - it’s a testament to the power of open-source innovation and the enduring impact of the Stable Diffusion legacy. The Black Forest Labs team’s ability to surpass established players like Midjourney and DALL-E 3 in critical areas such as text rendering and human figure accuracy demonstrates that technical excellence combined with community-focused development can challenge industry giants.

What makes FLUX particularly compelling is its dual approach: maintaining open-source accessibility through models like Schnell while offering premium capabilities through Pro variants. This strategy, coupled with strategic partnerships like NVIDIA and continuous innovation (4-megapixel support, multimodal inputs), positions FLUX at the forefront of democratizing high-quality AI image generation. As hardware requirements become more accessible and the Japanese community develops specialized workflows, FLUX is poised to shape the future of creative AI tools globally.