Black Forest Labs
Black Forest Labs is an independent visual AI company founded in 2024. Its headquarters are in Freiburg im Breisgau, Germany, and it also operates a lab in San Francisco. Co-founder Robin Rombach is CEO.
Company & leadership
- Current CEO
- Robin Rombach, co-founder and CEO.
- German legal entity
- BFL GmbH; Amtsgericht München, HRB 291820.
- Managing directors
- Robin Rombach, Andreas Blattmann and Axel Sauer.
- Ownership and funding
- Privately funded independent company; Series B led by AMP and Salesforce Ventures.
Founders
- Robin Rombach, Andreas Blattmann, Jonas Julius Müller, Sumith Kulal, Tim Dockhorn, Axel Sauer, Dominik Lorenz, Patrick Esser, Frederic Boesel and Harry Saini.
Founding and research background
The company launched publicly on 1 August 2024 with the FLUX.1 image model family. Its founding researchers had worked on methods and models including latent diffusion, Stable Diffusion and image-generation distillation. The launch also announced a $31 million seed round led by Andreessen Horowitz.
Investor Earlybird lists ten founders, rather than only the three names often used in short company descriptions. Its company profile records Freiburg im Breisgau as the headquarters and 2024 as the founding year.
Leadership and company locations
Black Forest Labs identified Robin Rombach as co-founder and CEO in its June 2026 G7 statement. The current German imprint separately lists Rombach, Andreas Blattmann and Axel Sauer as managing directors of BFL GmbH. These are distinct company titles.
The company describes its research operations as based in Freiburg and San Francisco. Its German registered address is Ingeborg-Krummer-Schroth-Straße 18, 79106 Freiburg im Breisgau.
Funding and ownership
In December 2025, Black Forest Labs announced a $300 million Series B led by AMP and Salesforce Ventures at a $3.25 billion post-money valuation. It also disclosed a prior Series A led by Andreessen Horowitz. The announcement put total funding above $450 million. These are dated financing figures, not a present valuation estimate.
Model development milestones
FLUX 3 was introduced in July 2026 as a model trained jointly on images, video and audio. This expanded the company's work beyond still-image generation. Its action releases also apply visual prediction to robot control.
The October 2026 Image Klein release added a compact downloadable image generator and an undistilled base for adaptation. Its Kommunity terms differ from both Apache-licensed FLUX releases and the earlier non-commercial model licenses.
Catalog scope
Practical guidance. The guides cover the principal public FLUX families, distinct base checkpoints and named generative tools. Quantized packages, file-format conversions and small decoder components are grouped with their parent model. The catalog distinguishes BFL endpoints from the FLUX 3 Max model post-trained and served by fal.