fal Research and Black Forest Labs Unveil High-Speed Video Model FLUX 3 Max
fal Research and Black Forest Labs have launched FLUX 3 Max, an image-to-video model that generates 10-second 720p clips in approximately 12 seconds.
Image source: fal Research / @fal
On October 20, 2026, fal Research announced the release of FLUX 3 Max, an advanced post-trained video generation model developed in partnership with Black Forest Labs (@bfl_ai). Confirmed publicly by @fal, the launch addresses two primary friction points in media generation: output fidelity and inference turnaround time.
Post-Training and Diffusion Transformer Architecture
fal Research explained that FLUX 3 Max joins its H3 Max system at the frontier of video synthesis. The release emphasizes what can be accomplished when foundation models are paired with specialized post-training routines and reinforcement learning infrastructure specifically built for video diffusion transformers. Rather than distributing raw foundation weights, this approach concentrates on refining frame-to-frame coherence and motion realism through targeted training, optimizing compute resource allocation while maintaining fine visual detail across complex motion sequences.
Benchmark Standing and High-Speed Generation Metrics
In performance evaluations, fal confirmed that FLUX 3 Max is tied for first place in image-to-video quality on independent third-party benchmarks. In addition to top-tier fidelity, the model achieves notable inference efficiency, generating a 10-second video at 720p or 768p resolution in approximately 12 seconds. Operating at near-real-time speeds represents a practical milestone for creative workflows, enabling designers and developers to iterate on visual concepts rapidly without enduring long rendering backlogs.
Promotional Pricing and Operational Considerations
To encourage adoption, fal is offering a 75% discount on FLUX 3 Max for the next three weeks, framing it as the most affordable frontier video option in its quality class. However, several operational caveats should be considered. The announcement did not identify the specific independent benchmarking organization or provide detailed scoring methodologies. Furthermore, the 75% discount is temporary, and standard rates after the promotional period remain unannounced, necessitating ongoing monitoring of official platform updates.