Editly

Veo 3 Fast, not full Veo 3 — and exactly what that means

Editly generates video on the Veo 3 Fast tier of Google's Veo 3 family: the quicker, cheaper sibling that keeps the same published clip-length ceiling, the same 16:9 and 9:16 framing and the same natively generated soundtrack, and trades away the top-end fidelity of the full model. Below are the specifications Google publishes, what one clip costs here, and which doors to Veo 3 Fast are still open.

Start creating

Veo 3 Fast vs Veo 3 vs Veo 3.1 Fast

Three tiers of one family. Google's own model cards give them near-identical technical specifications, so the honest differences are price, the editing controls added in 3.1, and which endpoint is still alive.

Veo 3
Veo 3 FastFast
Veo 3.1 Fast
Clip length the model supports4, 6 or 8 seconds4, 6 or 8 seconds4, 6 or 8 seconds
Output resolution720p or 1080p720p or 1080p720p or 1080p
Aspect ratios16:9 or 9:1616:9 or 9:1616:9 or 9:16
Frame rate24 fps24 fps24 fps
Sound generated with the pictureYesYesYes
Google list price, video with audio$0.40 per second$0.10 per second at 720p, $0.12 at 1080p$0.10 per second at 720p, $0.12 at 1080p
Google list price, silent video$0.20 per second$0.08 per second at 720p, $0.10 at 1080p$0.08 per second at 720p, $0.10 at 1080p
Image as the opening frameSupportedSupportedSupported, plus first-and-last-frame
Extending a clip, reference imagesNeitherNeitherBoth
Vertex AI endpointveo-3.0-generate-001, retired 30 June 2026veo-3.0-fast-generate-001, retired 30 June 2026veo-3.1-fast-generate-001, current

How to render a Veo 3 Fast clip here

  1. 01

    Pick Veo 3 in the model list

    Open the video generator and choose Veo 3 in the model list — that entry runs the Fast tier. The generator starts on Hailuo 2.3, so switch the model before you write anything, or the clip comes back from a different model entirely.

  2. 02

    Describe the shot, then describe the sound

    The prompt box takes up to 5,000 characters. Spend them on the frame — subject, lens, camera move, light — and then on what a microphone on set would pick up, because this model renders audio in the same pass rather than dubbing it afterwards. A line in quotation marks is read as dialogue.

  3. 03

    Choose the frame shape and the resolution

    Two aspect ratios, 16:9 for players and 9:16 for phones, and two resolutions, 720p or 1080p. Those are the model's own limits rather than ours. Pick the shape the clip will actually be published in; cropping a vertical cut out of a landscape render throws away most of the frame.

  4. 04

    Attach frame images if you have them

    The form accepts up to two images, each under 10 MB, to anchor the generation on artwork you already own — a product photo, a character sheet, a location plate. Leave the field empty for pure text-to-video.

  5. 05

    Generate, then download the MP4

    A run costs 330 credits at 720p and 360 credits at 1080p, deducted when the clip lands. Generation is queued, so the tab does not have to stay open. The file comes back as MP4 with the audio already mixed in.

What the Veo 3 Fast tier actually gives you

Sound that was never a separate job

Veo 3 was the first model in the line that could generate a soundtrack with the picture — traffic behind a city shot, birds in a park, lip-synced dialogue between characters — and the Fast tier inherits it. No scoring pass, no voice-over session, no sync drift.

A quarter of the price of the full model

On Google's own list, Veo 3 with audio is $0.40 per second of video while Veo 3 Fast is $0.10 at 720p. That ratio is the whole argument for the tier: four takes for the price of one, which matters far more while you are still finding the shot than the last few percent of fidelity does.

Text in, or a still you already own

Both modalities are supported. Start from a written shot description, or hand the model an image and let it move — the practical route when a product, a logo or a face has to look like itself rather than like a plausible invention.

Every frame carries SynthID

Google embeds its SynthID watermark in Veo output, imperceptibly, so a clip stays identifiable as AI-generated after it leaves you. Plan around it: platforms and clients that ask about provenance can detect it, and disclosing first is the easier path.

Where you can still reach Veo 3 Fast

This is the part most model pages leave out. The Veo 3 generation shipped in July 2025 and Google has moved on since; here is what each door actually opens onto today.

Gemini app
Google's APIs
Editly
Which model you getGoogle says Gemini Omni will replace Veo in the appVeo 3.1 and Veo 3.1 Lite; the Veo 3 endpoints were retiredThe Veo 3 Fast tier, unchanged
What it takes to startA paid Google AI subscriptionA cloud project, billing and codeAn email address
How you payMonthly plan; features vary by tier and regionPer second of generated video330 credits at 720p, 360 at 1080p, per clip
Controls in front of youA chat boxThe full request bodyPrompt, aspect ratio, resolution, two frame images
WatermarkSynthIDSynthIDSynthID

Veo 3 Fast, straight answers

Render on Veo 3 Fast without a Google subscription

One prompt, one aspect ratio, one resolution, and a clip that arrives with its own soundtrack. Credits come out of a single balance shared with every other model here.

Open the video generator