On August 1, 2024, Black Forest Labs announced both its existence and its first product, the FLUX.1 family of text-to-image models. The launch was notable because the founding team included researchers behind VQGAN, Latent Diffusion, and the Stable Diffusion models - the lineage that had powered the 2022 open text-to-image wave. Several founders had come from Stability AI, and the debut was widely read as that core team relaunching under its own banner, backed by a $31 million seed round led by Andreessen Horowitz.
FLUX.1 scaled a hybrid transformer-and-flow-matching architecture to 12 billion parameters and shipped in three tiers: a closed API-only [pro] model, an open-weight non-commercial [dev] model, and a fast Apache 2.0-licensed [schnell] model. On release it was widely judged to lead open image generation, especially on prompt adherence, rendering legible text inside images, and human anatomy. Within months FLUX.1 became a default open base model for image tools and fine-tunes, and the lab extended it into editing and video.
The launch marked a consolidation point in generative imagery: the talent that built Stable Diffusion regrouped into a focused lab, and its open-and-closed tiering set a template later startups followed.