Veo 3.1

Improved video generation model with higher fidelity, context-aware audio, and supports image references and frame interpolation.

Model Overview

A text-to-video generator that creates high-fidelity videos with context-aware audio. It builds on Veo 3 with improved quality.

Best At

• Generating high-quality videos from text descriptions.
• Maintaining subject consistency when using reference images.
• Smooth video transitions via last frame interpolation.
• Creating natural audio that matches the generated video content.

Limitations / Not Good At

• Reference images are limited to 16:9 aspect ratio and 8-second duration.
• Last frame option is omitted when reference images are used.
• Output is always a video file; no separate image or audio output.
• Specific input image resolutions required for different modes.

Ideal Use Cases

• Marketing and product demos with consistent branding.
• Social media video creation (short vertical or horizontal videos).
• Smooth transitions from images to video content.
• Videos with contextually relevant background vocals or sounds.

Input & Output Format

• Input: Text prompt (required) combined with optional parameters: aspect ratio, duration, starting image, last frame, reference images, negative prompt, resolution, and audio generation flag.
• Output: URI pointing to the generated MP4 video file.

Performance Notes

• High quality video generation requires substantial compute resources.
• Generation time may be longer than other modalities due to video synthesis.