Today we’re proud to announce that Hyperbolic and Black Forest Labs have partnered to bring FLUX.1 on Hyperbolic, marking a step toward building an open AI economy where everyone who contributes is rewarded. This comes weeks after Hyperbolic officially started supporting FLUX.1 [dev], an open-weight model developed by Black Forest Labs. Soon, Hyperbolic will extend its support to FLUX.1 [pro], offering even greater opportunities for developers to push what's possible with their model.
Who is Black Forest Labs?
Black Forest Labs (BFL) is a leading AI research group with deep roots in the generative AI community. Founded by a talented team of AI experts, BFL has developed foundational models like VQGAN, Latent Diffusion, and Stable Diffusion, including advanced iterations such as Stable Diffusion XL and Stable Video Diffusion. To date, BFL has raised $31 million in Series Seed funding led by Andreessen Horowitz, with participation from investors like Brendan Iribe, Michael Ovitz, Garry Tan, Timo Aila, and Vladlen Koltun, along with backing from General Catalyst and MätchVC.
What is FLUX.1?
BFL’s latest innovation, Flux.1, offers three distinct model variants:
FLUX.1 [pro] (coming soon to Hyperbolic) – A flagship model designed for professional, commercial use, delivering state-of-the-art performance in image generation with high-quality visual detail, prompt adherence, and output diversity.
FLUX.1 [dev] (available now on Hyperbolic) – An open-weight, guidance-distilled model for non-commercial applications. FLUX.1 [dev] retains much of the performance of FLUX.1 [pro] while offering greater efficiency. It’s already available on platforms like Hugging Face and Replicate, and now through Hyperbolic as well.
FLUX.1 [schnell] – A lightweight, fast model optimized for personal use, licensed under Apache 2.0 and available on GitHub.
FLUX.1 uses a hybrid architecture with multimodal and parallel diffusion transformer blocks optimized for performance and efficiency. It leverages flow matching, rotary positional embeddings, and parallel attention layers to achieve hardware efficiency, surpassing competitors like Midjourney v6.0, DALL·E 3, and Stable Diffusion 3.
Bringing Advanced Open Source Models to an Open Access AI Ecosystem
By supporting FLUX.1 [dev], Hyperbolic adds a powerful text-to-image tool to its open access AI ecosystem, aligned with its mission to make state-of-the-art AI services and products accessible to all. With FLUX.1 [dev] available, developers and researchers have access to a cutting-edge generative AI model for high-quality image synthesis.
Hyperbolic’s upcoming support for FLUX.1 [pro] will further enhance these capabilities, offering commercial users access to the most advanced features of Flux.1. This will open up new opportunities for enterprise-level applications, from content creation to AI-driven design and media production.
A Glimpse Into the Future
At Hyperbolic, we are building an open-access AICloud with the vision of creating an ecosystem where everyone—from model developers to compute providers—can benefit. Our partnership with BFL is a testament to our commitment to fostering an inclusive AI economy.
Stay tuned for more updates as Hyperbolic expands its support for world-class AI models for those at the edges of AI.
About Hyperbolic
Hyperbolic is a leading provider of open access AI computing and inference services, pioneering open access to AI for developers and researchers worldwide. With a mission to break down barriers that limit AI’s potential, Hyperbolic believes in a future where AI technology is universally accessible, empowering every individual and community with the tools to innovate, create, and advance our world. The Hyperbolic founding team is led by award-winning Math and AI researchers from UC Berkeley and the University of Washington.
Website | X | Discord | LinkedIn | YouTube | GitHub | Documentation