Model and Modality PlaybooksSeptember 15, 2026Flatkey Team

Veo 3 Pricing (October 2026): Veo 3 vs 3.1 Costs, API Access, and Plans

Veo 3 and Veo 3.1 cost $0.40 per second with audio on Vertex AI as of October 2026. Compare Veo 3.1 Fast and Lite, Gemini Omni Flash token pricing, Google AI plans, and what the Gemini API shutdown means for your code.

Veo 3 Pricing (October 2026): Veo 3 vs 3.1 Costs, API Access, and Plans

Veo 3 and Veo 3.1 cost $0.40 per second of 720p or 1080p video with native audio on Google Cloud Vertex AI, so a standard 8-second clip costs $3.20. Veo 3.1 Fast drops that to $0.10–$0.12 per second and Veo 3.1 Lite to $0.05–$0.08 per second, as of October 25, 2026. The bigger change is access: Google shut down the Veo 3.1 Preview models in the Gemini API on October 22, 2026, so developers now reach Veo through Vertex AI (the Gemini Enterprise Agent Platform) or switch to Gemini Omni Flash, which is billed per output token.

This guide separates the three ways people pay for Veo today: per second on Vertex AI, per token through Gemini Omni Flash in the Gemini API, and through Google AI subscription plans that include Flow and the Gemini app. Every number comes from an official Google page checked on October 25, 2026, and links to the source so you can re-check it before you budget a campaign.

Veo 3 pricing at a glance

All prices below are per second of generated video on Vertex AI generative AI pricing (checked October 25, 2026). "With audio" means Veo generates the soundtrack and dialogue together with the video.

Model	Vertex AI model ID	720p with audio	1080p with audio	4K with audio	720p/1080p video only
Veo 3.1	`veo-3.1-generate-001`	$0.40	$0.40	$0.60	$0.20 (4K: $0.40)
Veo 3.1 Fast	`veo-3.1-fast-generate-001`	$0.10	$0.12	$0.30	$0.08 / $0.10 (4K: $0.25)
Veo 3.1 Lite	`veo-3.1-lite-generate-001`	$0.05	$0.08	not offered	$0.03 / $0.05
Veo 3	`veo-3.0-generate-001`	$0.40	$0.40	not offered	$0.20
Veo 3 Fast	`veo-3.0-fast-generate-001`	$0.10	$0.12	not offered	$0.08 / $0.10
Veo 2	`veo-2.0-generate-001`	$0.50 (video only)	–	–	$0.50

Two things stand out. First, Veo 3 and Veo 3.1 have the same base price at 720p and 1080p, so there is no saving in staying on the older model. Second, turning off audio halves the standard Veo 3.1 price, which matters for B-roll, product loops, and anything you will score with your own music later.

What an 8-second clip costs

Most Veo 3.1 jobs are 4, 6, or 8 seconds long, per the Veo 3.1 model page on Vertex AI. Here is the per-clip math for the most common 8-second request:

Route	720p	1080p	4K
Veo 3.1 with audio	$3.20	$3.20	$4.80
Veo 3.1 video only	$1.60	$1.60	$3.20
Veo 3.1 Fast with audio	$0.80	$0.96	$2.40
Veo 3.1 Lite with audio	$0.40	$0.64	–
Gemini Omni Flash (Gemini API)	≈ $0.81	≈ $1.22	≈ $2.43

The Gemini Omni Flash row is calculated from token pricing, explained below. Remember that you pay for every generation, including the takes you throw away. If your team usually needs three attempts to get a usable shot, a "$3.20 clip" is really closer to $9.60 per finished shot, so draft in Fast or Lite and only render the final take in standard Veo 3.1.

Veo 3 vs Veo 3.1: what you are paying for

Veo 3 was the first Google model to generate synchronized audio, dialogue, and sound effects together with video. Veo 3.1 keeps that and adds better prompt adherence, reference images for character and product consistency, first-and-last-frame control, scene extension, and 4K output. On Vertex AI, both are general availability (GA) models.

	Veo 3	Veo 3.1
Native audio	Yes	Yes
Resolutions	720p, 1080p	720p, 1080p, 4K
Clip length	Up to 8 seconds	4, 6, or 8 seconds (reference-image mode: 8 seconds)
Price at 1080p with audio	$0.40/s	$0.40/s
Cheaper tiers	Veo 3 Fast	Veo 3.1 Fast, Veo 3.1 Lite
Gemini API status	Shut down June 30, 2026	Preview shut down October 22, 2026
Vertex AI status	GA	GA

For new work, there is little reason to choose Veo 3 over Veo 3.1. The price is the same and Veo 3.1 gives you more control. Keep Veo 3 only if you have a production pipeline already tuned to its output and you need to hold the look steady until a campaign ends.

The October 2026 access change: Veo leaves the Gemini API

According to the Gemini API deprecations page (checked October 25, 2026), Google retired Veo from the Gemini API in two steps:

veo-3.0-generate-001, veo-3.0-fast-generate-001, and veo-2.0-generate-001 shut down on June 30, 2026.
veo-3.1-generate-preview, veo-3.1-fast-generate-preview, and veo-3.1-lite-generate-preview shut down on October 22, 2026.

The recommended replacement in the Gemini API is Gemini Omni Flash (gemini-omni-1.1-flash), or the GA Veo models on the Gemini Enterprise Agent Platform, which is Vertex AI. The Gemini API video generation guide now points new users to Gemini Omni Flash. Veo prices have also been removed from the Gemini API pricing page, so any article that still quotes "Gemini API Veo pricing" is describing a route you can no longer call.

If you have code that calls a -preview Veo model ID, it will fail. You have two migration paths:

Stay on Veo: move the job to Vertex AI and use the -001 GA model IDs from the table above. You need a Google Cloud project with billing and a Cloud Storage bucket for outputs.
Stay on the Gemini API: switch to Gemini Omni Flash and budget by output tokens instead of seconds.

Gemini Omni Flash pricing compared with Veo

Gemini Omni Flash is a multimodal model that can generate video with conversational editing. On the Gemini API pricing page (checked October 25, 2026) it is paid-tier only: $1.50 per 1M input tokens, $9 per 1M text output tokens, and $17.50 per 1M video output tokens. Google publishes how many output tokens one second of video uses, which lets you convert it to a per-second price:

Resolution	Video output tokens per second	Approx. cost per second	Approx. 8-second clip
360p	1,931	$0.034	$0.27
720p	5,792	$0.101	$0.81
1080p	8,688	$0.152	$1.22
4K	17,376	$0.304	$2.43

That puts Gemini Omni Flash close to Veo 3.1 Fast on price, and well below standard Veo 3.1. Input tokens (your prompt and any reference images) are extra but are usually a small part of the bill. Treat these as estimates: the token count per second is Google's published figure, but your invoice is based on the tokens actually returned.

Google AI plans: the consumer way to use Veo

If you are a creator making a handful of clips a month, a subscription can be cheaper than the API. As of October 25, 2026, Google AI plans in the United States are:

Plan	Monthly price (US)	Video features relevant to Veo
Free	$0	Gemini app; limited Google Flow access
Google AI Plus	$4.99	Video generation in the Gemini app; 200 Google Flow credits
Google AI Pro	$19.99	4x Free usage limits; 1,000 Google Flow credits
Google AI Ultra	$99.99 or $199.99	5x or 20x higher usage limits than AI Pro

Google Flow is Google's AI filmmaking studio. Its credits are spent per generation, and the plan page now lists Gemini Omni Flash as the Flow video model, so check the in-app credit cost for each model before you plan a project. Subscriptions do not give you an API key: if you need to generate video from your own app or automation, you still need Vertex AI or the Gemini API.

Rule of thumb: subscriptions suit people who make clips by hand in Flow or the Gemini app. Per-second API pricing suits teams that generate video from code, need predictable unit costs per asset, or must keep usage records for clients.

How to call Veo 3.1 on Vertex AI

Veo on Vertex AI is a long-running job. You submit a request to predictLongRunning, get an operation name back, and poll fetchPredictOperation until the video is written to Cloud Storage. The request format follows Google's text-to-video guide for Vertex AI.

PROJECT_ID="your-project"
MODEL_ID="veo-3.1-fast-generate-001"

cat > request.json <<'JSON'
{
  "instances": [{ "prompt": "Slow dolly shot of a ceramic mug on a sunlit desk, steam rising, soft cafe ambience" }],
  "parameters": {
    "storageUri": "gs://your-bucket/veo/",
    "sampleCount": 1,
    "durationSeconds": 8,
    "resolution": "1080p",
    "generateAudio": true
  }
}
JSON

curl -X POST \
  -H "Authorization: Bearer $(gcloud auth print-access-token)" \
  -H "Content-Type: application/json" \
  -d @request.json \
  "https://us-central1-aiplatform.googleapis.com/v1/projects/${PROJECT_ID}/locations/us-central1/publishers/google/models/${MODEL_ID}:predictLongRunning"

A small Python helper keeps the cost visible before you submit anything:

# Veo prices per second on Vertex AI, checked October 25, 2026
PRICE = {
    ("veo-3.1-generate-001", "1080p", True): 0.40,
    ("veo-3.1-generate-001", "1080p", False): 0.20,
    ("veo-3.1-fast-generate-001", "1080p", True): 0.12,
    ("veo-3.1-lite-generate-001", "1080p", True): 0.08,
}

def clip_cost(model, resolution, audio, seconds=8, takes=1):
    return PRICE[(model, resolution, audio)] * seconds * takes

print(clip_cost("veo-3.1-generate-001", "1080p", True, takes=3))  # 9.6

Log the model ID, resolution, audio flag, and duration with every job. When finance asks why video spend doubled, those four fields usually explain it.

Ways to lower your Veo bill

Draft with Lite or Fast. Iterate on framing and timing at $0.05–$0.12 per second, then render the approved prompt once in standard Veo 3.1.
Turn off audio when you don't need it. Video-only output is half price on standard Veo 3.1.
Use the shortest clip that works. A 4-second clip costs half of an 8-second one, and many social cuts are shorter than 8 seconds anyway.
Avoid 4K until the final export. 4K is 50% more than 1080p on standard Veo 3.1 and 2.5x more on Veo 3.1 Fast.
Cap takes per asset. Set a budget such as three generations per shot, and send prompts that fail twice back for rewriting rather than re-rolling.

Using Veo alongside other video models with one key

Many teams now compare Veo with other video models instead of committing to one vendor. As of October 25, 2026, Flatkey does not list a Veo model in its public model catalog, so call Veo directly through Vertex AI. Flatkey does route other video models through one OpenAI-compatible /v1/videos endpoint with a shared balance. For example, the Seedance 2.0 model page lists 720p from $0.0789 per second and 1080p from $0.1942 per second, and Grok Imagine Video 1.5 lists $0.088 per second.

curl https://router.flatkey.ai/v1/videos \
  -H "Authorization: Bearer $FLATKEY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "seedance-2.0",
    "content": [{"type": "text", "text": "Slow dolly shot of a ceramic mug on a sunlit desk, steam rising"}],
    "resolution": "1080p",
    "ratio": "16:9",
    "duration": 5,
    "generate_audio": false
  }'

For a fuller side-by-side, see our Seedance vs Veo routing and pricing comparison, the broader AI video generation API pricing comparison, and the Seedance API pricing and setup guide. If you are planning a migration because a video model is being retired, our Sora 2 API deprecation watch covers the same checklist for OpenAI's video route.

FAQ

How much does Veo 3 cost per second?

On Vertex AI, Veo 3 and Veo 3.1 cost $0.40 per second at 720p or 1080p with audio, and $0.20 per second without audio, as of October 25, 2026. Veo 3.1 at 4K is $0.60 per second with audio.

How much is an 8-second Veo 3.1 video?

A standard 8-second Veo 3.1 clip with audio costs $3.20 at 720p or 1080p and $4.80 at 4K. The same clip on Veo 3.1 Fast costs $0.80 at 720p or $0.96 at 1080p, and on Veo 3.1 Lite it costs $0.40 at 720p.

Is Veo 3 still available in the Gemini API?

No. Veo 3 and Veo 3 Fast were shut down in the Gemini API on June 30, 2026, and the Veo 3.1 Preview models were shut down on October 22, 2026. Use the GA Veo models on Vertex AI, or switch to Gemini Omni Flash in the Gemini API.

Is Veo 3 free?

There is no free API tier for Veo. Google AI Plus ($4.99 per month) and higher plans include video generation in the Gemini app and Google Flow credits. New Google Cloud accounts may get trial credits, which can be spent on Vertex AI.

What is the cheapest way to use Veo 3.1?

For API use, Veo 3.1 Lite is the cheapest at $0.03 per second for 720p without audio, or $0.05 per second with audio. For occasional manual use, a Google AI plan with Flow credits can cost less than paying per second.