BlackForestLabs Unveils FLUX.3 Video Generation Model, Capable of 20-Second 1080p Videos with Audio

On August 5, BlackForestLabs announced the official launch of its FLUX3 video generation model. The model supports generating up to 20 seconds of 1080p-resolution video with native audio included. FLUX3 employs a unified architecture that jointly learns images, video, and audio, enabling it to output both video and audio in a single generation pass. It supports text-to-video, image-to-video, video-to-video, video-and-audio continuation from input clips, keyframe-to-video conversion, multilingual dialogue, and multi-shot sequencing. In text-to-video tasks, FLUX3 achieved a score of 1,135—outperforming both Gemini OmniFlash and Minimax H3. In image-to-video tasks, it scored 1,051, surpassing Seedance 2.0 and Minimax H3. Pricing for FLUX3 is based on the duration (in seconds) of the generated video. In Draft mode (fast/economical), only HD resolution is supported: text-to-video or image-to-video costs $0.06 per second, while video-to-video costs $0.12 per second. In standard HD mode, text-to-video or image-to-video costs $0.17 per second, and video-to-video costs $0.41 per second. In Full HD mode, text-to-video or image-to-video costs $0.29 per second, and video-to-video costs $0.53 per second.