Google· Video modelNewVeo 3.1 is the current version of Google's Veo 3. Write a prompt or add a start frame, turn on sound for dialogue and effects, and get a 4 to 8 second clip at up to 1080p. Credits show before you run.
Made with these models
Original clips made with this model. Read the prompt, or press Use this example to load the same settings.
Veo 3.1
Handheld selfie vlog in one continuous shot. She shouts to the camera: "Okay, it is touching down. We are leaving!" Behind her the funnel touches down and throws debris across the field, hail bounces off the truck and the camera shakes in the gusts as she turns to run. Sound: roaring wind, hail drumming on metal, a distant siren. No music, no text.
Veo 3.1
Slow push-in. The dog narrows its eyes and speaks firmly into the receiver with matching mouth movements: "No. Two treats by Friday, or no deal." Lightning flashes over the skyline behind it. Sound: rain on glass, distant thunder, creaking leather. No music, no text.
Veo 3.1Handheld selfie vlog in one continuous shot. She shouts to the camera: "Okay, it is touching down. We are leaving!" Behind her the funnel touches down and throws debris across the field, hail bounces off the truck and the camera shakes in the gusts as she turns to run. Sound: roaring wind, hail drumming on metal, a distant siren. No music, no text.
Veo 3.1Slow push-in. The dog narrows its eyes and speaks firmly into the receiver with matching mouth movements: "No. Two treats by Friday, or no deal." Lightning flashes over the skyline behind it. Sound: rain on glass, distant thunder, creaking leather. No music, no text.
Google built this update on Veo 3 with richer sound, closer prompt following and better image to video.
With sound on, speech, effects and room tone come with the picture, in sync. Put a line in quotes and a character says it. Sound is optional, so you pay for it only when needed.
Google says this version renders more true-to-life textures. You see it in skin, fabric, water and evening light, ideal for product shots, travel scenes and anything that should look filmed.
Add a start frame and the clip opens on your picture. Add an end frame and the model fills in the move between them, sound included: ideal for reveals and before and after shots.
Veo 3.1, Fast and Lite share one picker: test an idea on Lite for 5 credits, then render the final take on the full model.
All three make 4 to 8 second clips at up to 1080p, 16:9 or 9:16, with optional sound. The full model is the quality tier (about 2.5 minutes), Fast uses half the credits or less (about 1.5 minutes), and Lite costs the least (about 1 minute) but has no end frame.
| Model | This pageVeo 3.1 | Veo 3.1 Fast | Veo 3.1 Lite |
|---|---|---|---|
| Credits per clip | 33 (4 s, 720p) | 17 (4 s, 720p) | 5 (4 s, 720p) |
| Clip length | 4 to 8 s | 4 to 8 s | 4 to 8 s |
| Resolution | 720p, 1080p | 720p, 1080p | 720p, 1080p |
| Shapes (text to video) | 16:9, 9:16 | 16:9, 9:16 | 16:9, 9:16 |
| End frame | Yes | Yes | No |
| Sound | Optional sound, extra credits | Optional sound, extra credits | Optional sound, extra credits |
| Typical wait | About 2.5 min per clip | About 1.5 min per clip | About 1 min per clip |
Credits per clip at the default length and resolution, without optional sound. Longer, sharper or with-sound clips cost more; the button shows the credits before you run.
From idea to finished clip in three steps.
Describe the subject, action, camera and light. Add a start frame to animate a picture, or both frames for a transition.
Choose length, resolution, shape and sound, and make up to 4 clips per run. The credits on the button update as you go.
A Veo 3.1 clip takes about 2.5 minutes. Play it, download the MP4, or loop it with video to GIF.
Four tips drawn from Google's own prompt advice.
Google's guide lists what a good prompt holds: subject, action, style, camera position and move, and framing. Fit all five into one or two sentences, for example: close-up of an old fisherman mending a net on a pier at dawn, handheld camera slowly pushing in, warm documentary look. A vague prompt leaves the model guessing.
With sound on, write each spoken line in quotation marks and say who speaks it: the barista looks up and says, "Your usual?" Keep lines short, because an 8 second clip holds only one or two. English is the safest choice, as Google notes that other languages have not been fully evaluated.
Google's guide says the model follows cues for sound effects and ambient noise, so name them: rain on a tin roof, a slamming door, the murmur of a busy cafe, wind in the pines. If a clip needs no audio at all, switch sound off: it comes out silent and costs about half the credits.
For a start and end frame clip, use two pictures with the same shape, light and style, and describe only the move between them: the camera glides from the doorway to the window. Matching pictures give a smooth transition; very different ones can force a jump cut. Need the frames? Make them with text to image.
Short shots that need real light, believable motion or a voice.
Product shots and spoken ad hooks for the web or for reels.
Vertical clips with dialogue, ready to post without a voice-over.
Show a client a script beat they can see and hear.
Establishing shots and talking scenes for intros and courses.
Versions, credits, wait times, sound and commercial use.
Compare Veo 3 with other top video models, or turn an image to video with any of them.
ByteDance · Video modelNewSeedance 2.0ByteDance's flagship: film-like camera work, big action and physics, sound at no extra cost. Does not take photos of real people.62 credits per 5 s clip
Kuaishou · Video modelNewKling 3.0Kling's top tier in 1080p: smooth, controlled motion; start and end frames.23 credits per 5 s clip
xAI · Video modelNewGrok ImaginexAI's Grok video: lively motion, with sound.29 credits per 5 s clip
MiniMax · Video modelHailuo 2.3Strong push-ins and lively faces; takes a few minutes.12 credits per 6 s clipWrite a prompt, choose sound and see the credits before you run.