Black Forest Labs launches FLUX 3 model for video generation
Black Forest Labs has released the FLUX 3 AI model, capable of generating up to 20-second native 1080p videos with audio.
Black Forest Labs officially launched the FLUX 3 video generation model on August 5, according to a company announcement. The model is designed to produce videos up to 20 seconds in length at a native 1080p resolution, including original audio.
The company stated that FLUX 3's performance in text-to-video and image-to-video generation surpasses that of competing models such as Bytedance's Seedance 2.0 and Minimax's H3. In performance benchmarks, FLUX 3 achieved a score of 1,135 for text-to-video tasks and 1,051 for image-to-video tasks.
FLUX 3 supports a range of generation capabilities, including text-to-video, image-to-video, video-to-video, extending existing videos with audio, keyframe-to-video conversion, multilingual conversations, and multi-camera sequence synthesis. The model utilizes a unified architecture for joint learning of image, video, and audio.
Pricing for the model is based on the duration of the generated video. Costs vary depending on the generation mode and resolution, ranging from $0.06 per second for draft-mode text/image-to-video generation to $0.53 per second for full HD video-to-video generation.