Google’s Veo models have a reputation as the premium option in AI video, and part of that reputation is price. A single 8 second clip on the top tier of the current release can cost more than a minute of footage on some rivals. But the model is not one price. It is three tiers, and knowing which shots deserve which tier is the difference between a sensible budget and a painful one.
The three tiers, briefly
All three make clips at 720p or 1080p in 16:9 or 9:16, and all three generate sound with the picture.
Standard is the top quality tier, built for faces, dialogue and hero shots. It runs 4, 6 or 8 seconds at 12 credits a second with audio at list price, or 8 without. It is also the only tier that accepts up to three reference images to keep a subject or look consistent.
Fast runs the same lengths at 4.5 credits a second with audio, or 3 without. It handles motion and action well. In side by side runs of one dance prompt across tiers, Fast opens on movement and swings the fabric bigger, which suits social clips.
Lite costs 2 credits a second at 1080p and 1.5 at 720p, with audio always on. At 1080p it renders 8 second clips only, which works out to 16 credits a clip.
The gap in cost is large, but the gap in look is smaller than you might expect. Run the same prompt of a dancer in crimson silk under a single shaft of light on all three tiers and each keeps the staging, the haze and the red glowing against grey concrete. Standard gives the richest atmosphere and the cleanest silhouette. The cheaper tiers give you the same shot with a little less polish.
A 32 second spot, planned by tier
Say you are cutting a short ad for a coffee brand from four 8 second clips at 1080p. Putting all four on Standard costs 384 credits at list price. Here is a smarter split.
The opener, a slow flyover of a hillside farm at sunrise, is pure atmosphere. No faces, no speech. Put it on Lite for 16 credits.
The product shot, cream blooming through iced coffee while condensation beads on the glass, needs motion and texture. Fast handles liquid and light well, and with audio for the clink of ice it costs 36 credits.
The barista pulling a shot and pouring latte art is action with hands. Fast again, 36 credits.
The closer is a customer taking a sip, looking up and saying a line to camera. Faces and speech are where Standard earns its price: 96 credits.
Total: 184 credits instead of 384, and the expensive tier lands exactly where viewers look hardest.
Where the audio switch saves money
On Standard and Fast, audio can be switched off. If a clip will sit under a voiceover or music bed, turn it off: an 8 second Standard clip drops from 96 to 64 credits, and Fast drops from 36 to 24. Lite has no switch, but at its price that hardly matters.
Duration is the other lever. Credits scale with length, so a 4 second clip costs half as much as an 8 second one. Use 8 seconds when a camera move or a spoken line needs room, and 4 for inserts and cutaways.
Getting dialogue right
The prompting habits for Veo 3.1 are consistent across tiers. Open with shot size and camera move, then one specific subject doing one thing, then light, lens and a “Sound:” line. For speech, put the exact words in double quotes and describe the voice, such as “in a low, steady voice.” Finish with “No subtitles.” to keep text out of the frame.
One clear action per clip fixes most problems. One spin, one pour, one line. Ask for three things in 8 seconds and the model has to rush all of them.
A dialogue prompt worth adapting: “Medium close-up, slow push in. A tired night shift nurse in navy scrubs leans on a vending machine, looks at the camera and says quietly: ‘Third coffee. Don’t judge me.’ Cool fluorescent light, 50mm lens, shallow depth of field. Sound: vending machine hum, a distant pager. No subtitles.”
Using images to hold continuity
All three tiers take a start image, and you can add a last frame for the clip to land on. For a multi-clip spot, that is how you keep continuity: take the final frame of one clip and use it as the first frame of the next. Standard’s reference images are the stronger option for keeping a character consistent, though they run at 16:9 and 8 seconds and take the place of a last frame.
The honest downsides
Eight seconds is the maximum per clip, so anything longer is an editing job. Aspect ratios are limited to widescreen and vertical, with no square option. And Standard is expensive enough that iterating on it blind is a bad idea. Test composition on Fast or Lite, then move the final prompt up a tier.
Veo 3.1 sits in PixelDojo plans alongside the other major video models, which makes this kind of tier mixing easy to manage from one account. Treat Standard as your close-up lens, Fast as your workhorse and Lite as your b roll camera, and a premium model becomes far more affordable.