Veo 3.1 Text-to-Video is Google's video model. It generates a clip with native audio from a text prompt alone, with no source image. Generation runs in the background, so the clip appears when it is ready rather than immediately.
What it's good at
Veo 3.1 turns a written description straight into moving footage, with synchronised sound and no source image. It suits:
- Establishing and b-roll shots: a usable clip from a prompt, with no shoot and no source footage.
- Short narrative or product clips: a short clip carries a single beat or a product turn.
- Stylised or photoreal scenes: the prompt sets the look, so you are not limited to what you can film.
Because audio is native, a single prompt can carry both the action and the sound of the scene.
Veo 3.1 works from a prompt alone. If you need to animate an image you already have, use Veo 3.1 Image-to-Video.
Tiers
Veo 3.1 comes in three tiers, chosen per generation from the Tier pill. The tier changes quality, latency, and cost. It does not change your prompt or the settings below.
- Standard: the highest-quality tier, with native audio and output up to 4K. Choose it for the final render, or when the clip has to look its best.
- Fast: the same model tuned for lower latency and cost, still with native audio. Choose it while you are still iterating on a shot.
- Lite: the lowest-cost tier, and it tops out at 1080p with no 4K. Choose it for quick drafts and high-volume passes.
Settings
- Resolution: sets the frame size of the clip. Choose 720p, 1080p, or 4K on Standard and Fast. Lite offers 720p or 1080p. Default: 720p. Higher resolutions cost more.
- Duration: sets the length of the clip. Choose 4, 6, or 8 seconds from a fixed set rather than a free slider. Default: 4.
Practical tips
- Describe motion and camera, not just the scene. Motion is what the model has to animate.
- Keep the prompt to one clear action or shot. Busy, multi-event prompts are harder to render cleanly.
- Draft on Lite or Fast to settle the framing and the prompt wording. Then re-run the prompt on Standard. Expect a different clip, because there is no seed to hold.
- Mention the sound you want. Native audio means the prompt can shape it.
- The prompt box holds up to 2,500 characters. Trim wording rather than stacking clauses as you approach it.
