Veo 3.1 fast

Veo 3.1 Fast is positioned as a streamlined text-to-video model in Google's Veo family, designed for fast turnaround rather than maximum quality. It accepts natural language prompts as its primary input and supports both 16:9 and 9:16 aspect ratios, defaulting to widescreen 16:9. Generated clips can be 4, 6, or 8 seconds long, with 8 seconds as the default, and the model is documented to optionally include generated audio alongside the video, making it suitable for short-form social content, rapid prototyping of visual ideas, and ad or storyboard previews where speed matters more than maximum fidelity.

Beyond pure text prompting, Veo 3.1 Fast offers image-conditioned and frame-conditioned workflows. An optional reference image can be supplied at 1280x720 (landscape) or 720x1280 (portrait) to anchor the visual style of the generated clip, and an additional last-frame image can be uploaded to drive interpolation, effectively creating a transition between the two provided images. The model is surfaced on Replicate under the google/veo-3.1-fast path, exposing the same prompt, image, and last_frame inputs through a simple API, which gives developers an accessible integration point for building video-from-text, image-to-video, and keyframe-to-keyframe pipelines without managing the underlying infrastructure.