Google Veo 3.1: How to Use It (Credits and Prompts)
Google Veo 3.1 on VdoBloom: 40 credits on Fast or 160 on High Quality per clip, no duration selector, three input modes, and prompts that pass.
Google Veo 3.1 runs on VdoBloom as two tiers — Fast at a flat 40 credits per clip and High Quality at a flat 160 — and there is no duration selector: every generation returns one fixed-length clip with native audio. You choose between 16:9 and 9:16 only, and you can work from text, from a first and last frame, or from three reference images on the Fast tier.
Updated October 2026.
Disclosure: VdoBloom is our platform.
What Veo 3.1 is
Veo 3.1 is Google’s video model, added to VdoBloom’s video tools in October 2025 and listed simply as “Veo 3.1” — there is no separate older Veo entry to confuse it with. Two tiers sit under that name in the picker: Fast, badged as the fastest option, and High Quality. Both are the same underlying model at different quality settings, and VdoBloom’s model picker scores the family 4.6.
Its defining trait, compared with Kling or Runway in the same picker, is that length is not a setting. You do not ask Veo for 5 or 10 seconds; you write a prompt and one clip comes back. That sounds limiting and, for storyboarded work, it is — but it also removes a whole class of decisions, and it makes budgeting trivial, because the price of a Veo generation is a fixed number rather than a duration multiplied by a rate.
What Veo does bring is native audio. Dialogue, ambience and effects are produced with the picture in the same pass, so a clip arrives with sound attached rather than needing a separate audio step.
What it is good at
- Clipped, self-contained shots. A single action in a single setting — a door opening, a car pulling away, a person turning to camera — is exactly what a fixed-length generation handles best.
- Sound with the picture. Native audio means spoken lines and environmental sound come back with the frame.
- Planned transitions. Supplying a first and a last frame lets you define where the motion starts and where it must land.
- Guided generation from references. On the Fast tier, three reference images steer the look of a 16:9 clip.
- Non-English prompts. An Auto-Translate toggle converts a prompt written in another language before generation, and it is on by default.
- Predictable cost. Two prices exist, 40 and 160 credits, and neither changes with what you ask for.
Credits and input modes
| Tier | Cost per generation | Input modes | Aspect ratio |
|---|---|---|---|
| Veo 3.1 Fast | 40 credits, flat | Text to Video; Frame Transition (2 images, first and last frame); Reference Mode (3 images) | 16:9 or 9:16 |
| Veo 3.1 High Quality | 160 credits, flat | Text to Video; Frame Transition (2 images) | 16:9 or 9:16 |
Three details follow from that table. Reference Mode is locked to 16:9 and to the Fast tier, so a three-image reference job on High Quality is not available. There is no square 1:1 option at either tier, which most other video models on the platform do offer. And because price is flat, a High Quality render costs four times a Fast one regardless of what happens inside the clip.
The practical consequence is a two-stage workflow that the cost structure invites: draft on Fast for 40 credits, and only when the prompt is doing what you want, re-run it on High Quality for a final delivery. A Fast clip at 40 credits is also a reasonable test when you have 10 free signup credits plus a plan or pack behind it — pricing covers Lite at $15 a month for 300 credits (about $0.05 a credit), $10 a month billed yearly, and one-time packs from $10.39 for 200 credits that never expire.
Prompts that work with a fixed-length clip
Veo accepts prompts up to 5,000 characters. Because you cannot choose the length, the prompt has to imply one: describe a single action that resolves, not a sequence of beats.
- One action, one setting. “A woman in a wool coat steps out of a shop doorway into light rain, opens her umbrella and walks to the right out of frame. Static camera at eye level, overcast light, wet pavement reflecting the shop sign. Sound: rain on fabric, a bell on the shop door, distant traffic.”
- With a spoken line. “Medium shot of a mechanic in overalls leaning on a workbench, wiping his hands on a cloth. He looks up and says: we can have it done by Friday. Warm hanging bulbs, cluttered garage behind him, handheld camera drifting slightly. Sound: a radio playing in the background, tools being set down.”
- Frame transition. “Start on the supplied first frame, a closed blue door in a brick wall. End on the supplied last frame, the same doorway open onto a bright garden. Fill the motion: the door swings inward, light spills onto the step, leaves stir. Camera locked off. Sound: a latch releasing, birds beyond the door.”
- Reference mode, 16:9 (Fast tier). “Use the three reference images for the palette, the prop and the location. A hand lifts the ceramic cup from the table, steam curling, then sets it down. Slow push in, soft window light from the left. Sound: a cup on ceramic, room tone, a chair creaking.”
- Product beat. “A pair of trainers rotates slowly on a black plinth as a hard key light sweeps across the fabric. Tight framing, shallow depth of field, seamless black background. Sound: a low hum and one soft impact as the rotation stops.”
- Narrative moment. “A boy crouches at the edge of a pond and drops a flat stone. Ripples spread outward. Low camera near the water, late afternoon light behind him. Sound: water, insects, a dog barking far away, no music.”
Limits and the strict filter
- No duration control. Every generation returns one fixed-length clip. If you need an exact 6 or 12 seconds, choose a different family.
- Aspect ratios: 16:9 and 9:16 only. No square output.
- Prompt limit: 5,000 characters. An Auto-Translate toggle handles non-English prompts and is on by default.
- Reference Mode is Fast-only and 16:9-locked, with three images.
- Strict content filtering. Veo is labelled a strict-filtering model on VdoBloom, and Google’s provider-side rules apply in full. Clothed, non-provocative, adult content in ordinary settings passes; anything near the line is likely to be refused. Failed generations have their credits returned, so a refusal costs you time rather than money.
- Real faces need consent. Only your own photo, or one where the person agreed, should be used in frame-transition or reference modes.
Filtering is the most common reason a Veo prompt comes back empty, and it usually has a specific cause — a real person’s likeness, a brand, a prop that reads as a weapon, or wording that sounds like something you are not asking for. The guide to Veo refusing prompts works through those cases one at a time.
Frequently asked questions
How much does Veo 3.1 cost on VdoBloom?
40 credits per generation on Fast and 160 on High Quality, flat and regardless of input mode or content. There is no per-second rate to calculate, and the price is shown before you generate.
How long are Veo 3.1 clips?
You cannot set the length. Every generation produces one fixed-length clip, which is why the price is flat rather than duration-based. For controlled lengths of 3 to 15 seconds, a model with a duration selector is the better fit.
Can Veo 3.1 take reference images?
Yes, in two ways. Frame Transition takes two images — a first and a last frame — on both tiers. Reference Mode takes three images to guide the look, but is available only on Fast and only at 16:9.
Why is my Veo prompt refused?
Almost always the strict content filter. Veo is one of the more tightly moderated models on the platform: real-person likeness, brand references, weapon-like props and suggestive phrasing are the usual triggers. Rewrite literally, drop names and logos, and pick a model with looser filtering if the content is legitimate but outside Veo’s rules.
Should I use Fast or High Quality?
Fast for anything you have not yet proved: 40 credits per attempt makes iteration cheap, and it is also the only tier with Reference Mode. High Quality, at 160 credits, is best reserved for a final render of a prompt that already works on Fast.
Open Text to Video, select Veo 3.1, leave the tier on Fast and write a prompt that resolves in one action and one setting. Once the clip does what you want, switch the same prompt to High Quality for the delivery pass. If you are comparing families, Veo vs Kling vs Runway sets the three side by side, and how video credits work explains how per-second models differ from Veo’s flat rate.
Ready to try it?
Create your first AI video in minutes — no credit card required.
Start Creating Free →