Research
Explore our research articles and technical deep dives into the latest developments in generative AI and visual intelligence.
FLUX 3 Action: A 7B World Action Model for Robot Control
An open-weight 7B world action model built on the FLUX 3 backbone, setting a new state of the art on RoboLab-120 at less than half the parameters of the previous best open model.
Self-Supervised Flow Matching for Scalable Multi-Modal Synthesis
A modality-agnostic, self-supervised flow matching framework for image, video, and audio generation.
FLUX.2: Analyzing and Enhancing the Latent Space of FLUX – Representation Comparison
A comprehensive study comparing different autoencoder representations for diffusion models, including SD-VAE, RAE, FLUX.1 and FLUX.2.
FLUX.1 Kontext: Flow Matching for In-Context Image Generation and Editing in Latent Space
FLUX.1 Kontext is a unified model for image editing and generation in context, achieving fast, state-of-the-art results with strong consistency and quality.