---
title: "Veo 3.1 API vs Google Cloud Vertex AI: Honest Comparison | imaginary.app"
description: "Google Cloud Vertex AI vs the imaginary.app API for Veo 3.1: setup, auth, features, real per-clip prices (Google list is lower) and when each fits best."
url: https://www.imaginary.app/blog/veo-3-api-vs-google-vertex-ai
---

Updated November 2026

# Veo 3.1 API vs Google Cloud Vertex AI: An Honest Comparison

Google Cloud (Vertex AI, now "Gemini Enterprise Agent Platform") is the cheaper way to call Veo 3.1 per clip, and it exposes more features. imaginary.app costs about 1.6–3.1× Google's list price per clip. In return you get no Cloud project, service account or quota setup, one static key, prepaid packs and automatic refunds when a job fails.

Key takeaways

- Same models. We call Google's `veo-3.1-*-001` IDs on Google Cloud: Veo 3.1 and Veo 3.1 Fast are generally available there; Veo 3.1 Lite is in Preview.
- **Google is cheaper per clip.** For example, Fast 1080p with audio for 8 seconds is $0.96 at Google list price vs $2.13–$2.50 here.
- Google exposes more: extend, reference images and up to 4 videos per request. Our API does text, image and first-and-last-frame, one clip at a time.
- We are simpler. A `veo_` key and two endpoints, vs a GCP project, billing, IAM, OAuth tokens and a bucket.
- Veo 3 is retired on both. The Gemini API Veo previews shut down on October 22, 2026.

Veo 3.1 without Google Cloud setup

Skip the Cloud project, keep the model

One API key, prepaid credits, automatic refunds on failed generations.

[Get API Access](https://www.imaginary.app/api)[Read the Docs](https://www.imaginary.app/api-docs)

Sign in with Google to create an account; buy credits only when you need them.

## Where Veo 3.1 lives in late 2026

Google renamed Vertex AI to "Gemini Enterprise Agent Platform". The docs moved, but the REST host (`aiplatform.googleapis.com`) and the request shape did not change. Veo 3 (`veo-3.0-generate-001`) was retired on June 30, 2026. Veo 3.1 and Veo 3.1 Fast are generally available, and Veo 3.1 Lite is in Preview on Cloud. The Gemini API (AI Studio keys) dropped Veo entirely on October 22, 2026, which is why many developers now have to choose between setting up Cloud and using a third-party key. If you were on the Gemini API, start with the [Gemini API shutdown guide](https://www.imaginary.app/guides/veo-3-1-gemini-api-shutdown).

## Side by side

|                     | Google Cloud (Vertex AI)                                                            | imaginary.app API                                            |
| ------------------- | ----------------------------------------------------------------------------------- | ------------------------------------------------------------ |
| Setup               | GCP project, billing account, Vertex AI API enabled, IAM permissions, output bucket | Sign in with Google, create a key                            |
| Auth                | Short-lived OAuth access tokens (service account or user credentials)               | Static veo\_ key in a header                                 |
| Billing             | Per second of output on your Cloud invoice                                          | Prepaid credits per clip; packs from $9.99 or a monthly plan |
| Models              | Veo 3.1, 3.1 Fast (GA), 3.1 Lite (Preview)                                          | The same three, via veo3-quality, veo3-fast, veo3-lite       |
| Inputs              | Text, image, first/last frame, reference images, video extend                       | Text, image, first/last frame                                |
| Videos per request  | 1–4                                                                                 | 1                                                            |
| Output              | Written to your Cloud Storage bucket                                                | Public MP4 URL returned in the status response               |
| Failed generations  | Check Google's billing terms for your account                                       | Credits refunded automatically                               |
| Region              | us-central1 for Veo 3.1                                                             | Handled for you                                              |
| Enterprise controls | IAM, organization policy, your own agreement with Google                            | Per-key permissions and up to 10 keys per account            |

## Price per 8-second clip

Google lists Veo per "count" without defining the unit. We read it as one second of output, the usual way per-second video prices are quoted, and multiply by 8. Our range runs from our cheapest credit (largest monthly plan) to our most expensive (smallest one-time pack).

| 8 s clip              | Google Cloud list | imaginary.app            | Ratio    |
| --------------------- | ----------------- | ------------------------ | -------- |
| Lite 720p, audio      | $0.40             | 3 credits = $0.64–$0.75  | 1.6–1.9× |
| Lite 1080p, audio     | $0.64             | 5 credits = $1.07–$1.25  | 1.7–2.0× |
| Fast 720p, audio      | $0.80             | 10 credits = $2.13–$2.50 | 2.7–3.1× |
| Fast 1080p, audio     | $0.96             | 10 credits = $2.13–$2.50 | 2.2–2.6× |
| Fast 1080p, silent    | $0.80             | 6 credits = $1.28–$1.50  | 1.6–1.9× |
| Quality 1080p, audio  | $3.20             | 26 credits = $5.55–$6.49 | 1.7–2.0× |
| Quality 1080p, silent | $1.60             | 13 credits = $2.77–$3.25 | 1.7–2.0× |
| Quality 4K, audio     | $4.80             | 38 credits = $8.11–$9.49 | 1.7–2.0× |

Google also offers lower rates through savings plans, so the gap can be wider than shown. One thing to check before you budget: at our last check, Google's Veo 3.1 model page listed pay-as-you-go as "Not supported" for the 3.1 models, while the pricing page shows per-second rates. Ask Google which consumption option applies to your project. To price your own mix of durations and resolutions, use the [Veo 3.1 cost calculator](https://www.imaginary.app/tools/veo-3-1-cost-calculator).

## Is the Cloud setup worth it at your volume?

Since the Gemini API Veo previews shut down, Google Cloud is the only official API for Veo 3.1. Before your first clip there, someone on your team has to:

1. Create a Google Cloud project and link a billing account.
2. Enable the Vertex AI API in that project.
3. Create a service account (or set up workload identity) and give it permission to call Vertex AI.
4. Create a Cloud Storage bucket for output and let the model write to it.
5. Write code that mints and refreshes OAuth access tokens, calls the `us-central1` endpoint and polls the operation.
6. Check your Veo quota, and request more if your launch needs it.
7. Set budget alerts, because Cloud bills after use, not before.

None of this is hard for a team already on GCP. For a solo developer or a team testing one feature, it is a day of work and a new system to look after. Whether it pays off depends on the per-clip gap multiplied by your volume:

### Veo 3.1 Fast, 1080p with audio, 8 s (the default veo3-fast request)

| Clips per month | Google Cloud list | imaginary.app | Extra you pay us |
| --------------- | ----------------- | ------------- | ---------------- |
| 10 clips        | $9.60             | $21.33–$24.98 | $11.73–$15.38    |
| 50 clips        | $48.00            | $107–$125     | $58.65–$76.88    |
| 200 clips       | $192              | $427–$500     | $235–$308        |
| 1,000 clips     | $960              | $2,133–$2,498 | $1,173–$1,538    |

### Veo 3.1 Lite, 720p with audio, 8 s

| Clips per month | Google Cloud list | imaginary.app | Extra you pay us |
| --------------- | ----------------- | ------------- | ---------------- |
| 10 clips        | $4.00             | $6.40–$7.49   | $2.40–$3.49      |
| 50 clips        | $20.00            | $32.00–$37.46 | $12.00–$17.46    |
| 200 clips       | $80.00            | $128–$150     | $47.98–$69.85    |
| 1,000 clips     | $400              | $640–$749     | $240–$349        |

Compare the right-hand column with what Cloud costs you to set up and run. As an illustration only, assume 8 hours of setup and 1 hour a month of upkeep at $100.00 an hour. Cloud then pays for itself in the first month at roughly 586–768 Fast clips, and below about 86 clips a month the saving doesn't cover the upkeep. Put in your own numbers; Google's savings plans lower its side further.

## The same request on both

Both APIs are asynchronous. Google returns a long-running operation that you poll with `fetchPredictOperation`. We return a `taskId` that you poll with a GET. The Google example is a trimmed-down version of the request our own backend sends for every Fast job:

### Google Cloud

```
# Prerequisites: a Google Cloud project with billing, the Vertex AI API enabled,
# a principal with permission to call it, and a Cloud Storage bucket for output.
PROJECT_ID="your-project"
TOKEN="$(gcloud auth print-access-token)"   # short-lived OAuth token, not an API key
MODEL="veo-3.1-fast-generate-001"
BASE="https://us-central1-aiplatform.googleapis.com/v1/projects/$PROJECT_ID/locations/us-central1/publishers/google/models/$MODEL"

# 1. Start a long-running operation
curl -X POST "$BASE:predictLongRunning" \
  -H "Authorization: Bearer $TOKEN" -H "Content-Type: application/json" \
  -d '{
    "instances": [{ "prompt": "A lighthouse on a cliff at dusk, waves crashing" }],
    "parameters": {
      "aspectRatio": "16:9", "durationSeconds": 8, "resolution": "1080p",
      "generateAudio": true, "sampleCount": 1,
      "storageUri": "gs://your-bucket/veo-output/"
    }
  }'
# -> { "name": "projects/.../operations/OPERATION_ID" }

# 2. Poll the operation
curl -X POST "$BASE:fetchPredictOperation" \
  -H "Authorization: Bearer $TOKEN" -H "Content-Type: application/json" \
  -d '{ "operationName": "projects/.../operations/OPERATION_ID" }'
```

### imaginary.app

```
# Prerequisites: an imaginary.app account with credits and a veo_ API key.

# 1. Start a job
curl -X POST https://api.veo3gen.app/api/generate \
  -H "Authorization: Bearer $IMAGINARY_API_KEY" -H "Content-Type: application/json" \
  -d '{
    "model": "veo3-fast",
    "prompt": "A lighthouse on a cliff at dusk, waves crashing",
    "audio": true,
    "options": { "aspectRatio": "16:9", "duration": 8, "resolution": "1080p" }
  }'
# -> 202 { "taskId": "veo_...", "creditsRequired": 10, ... }

# 2. Poll the job
curl -H "Authorization: Bearer $IMAGINARY_API_KEY" \
  https://api.veo3gen.app/api/status/veo_...
# -> { "status": "completed", "result": { "videoUrl": "https://...mp4" }, ... }
```

The difference that matters in production is authentication. On Cloud, your server mints OAuth tokens from a service account and refreshes them about every hour. With us, it sends the same key on every call. `https://api.imaginary.app` is an alias for the same API and accepts the same key. For the full parameter list, see [Veo 3.1 API access](https://www.imaginary.app/blog/veo-3-1-api-access) or the [API docs](https://www.imaginary.app/api-docs).

## Choose Google Cloud when…

- You generate enough video that the per-clip difference adds up to more than your setup and maintenance time.
- You already run on GCP and your security team prefers that data stays inside your own Cloud project and bucket.
- You need extend, reference images or several samples per request.
- You need a contract directly with Google, or committed-use discounts.

## Choose imaginary.app when…

- You want Veo 3.1 working today without a Cloud project, IAM roles or quota requests.
- Your volume is low or unpredictable, and prepaid packs fit better than a Cloud billing account.
- You were on the Gemini API previews and need a key-based replacement for Veo. Expect to change your request code: our request and response shape differs from Gemini's.
- You want failed and filtered generations refunded without filing a ticket.

| Workload                                             | Typical shape                                  | Better fit                      |
| ---------------------------------------------------- | ---------------------------------------------- | ------------------------------- |
| Prototype or demo                                    | Dozens of clips, deadline in days              | A key-based API                 |
| Agency client spots                                  | Bursty; a few campaigns a month; often no code | Web app or key-based API        |
| Indie SaaS feature (first months)                    | Usage unknown; must not block launch           | Key-based API; revisit at scale |
| Production pipeline                                  | Thousands of clips a month, predictable        | Google Cloud                    |
| Regulated or GCP-native company                      | Data must stay in your own project             | Google Cloud                    |
| Needs extend, references or several samples per call | Feature-driven                                 | Google Cloud                    |

Many teams do both: start with a key-based API to prove demand, then move the steady high-volume path to Cloud once the numbers justify it. Both are submit-poll-download, so the move is a client swap. The [integration architecture guide](https://www.imaginary.app/blog/integrate-ai-video-generation-in-your-app) shows how to keep the provider behind one interface.

Honest bottom line

If price per second is your only criterion and you can run GCP, go to Google directly. We are a convenience layer on top of Google, not a discount on it.

## Frequently asked questions

### Is imaginary.app cheaper than Vertex AI for Veo 3.1?

No, not per clip. At Google Cloud's list price, an 8-second clip costs less than through us. On the cases in our table we cost about 1.6 to 3.1 times Google's list price. What you get for the difference is no Cloud setup, a static API key, prepaid packs and automatic refunds on failed jobs.

### Is Vertex AI still called Vertex AI?

Google now brands it "Gemini Enterprise Agent Platform". The API host is still aiplatform.googleapis.com, and Veo 3.1 runs there under the model IDs veo-3.1-generate-001, veo-3.1-fast-generate-001 and veo-3.1-lite-generate-001.

### Is Veo 3 still available on Vertex AI?

No. Google retired veo-3.0-generate-001 and veo-3.0-fast-generate-001 on June 30, 2026. Veo 3.1 replaced them. Veo 3.1 and Veo 3.1 Fast are generally available on Google Cloud, which lists them as supported until “May 4, 2027 or later”; Veo 3.1 Lite is in Preview.

### Are the videos the same quality?

We call the same Google Cloud model IDs, so the model is the same. Our API exposes fewer of its options: no extend, no reference images and one video per request.

### How many videos a month before Google Cloud is worth it?

There is no single number. Multiply your monthly clip count by the per-clip difference (for an 8-second Fast 1080p clip with audio, about $1.17 to $1.54) and compare it with what setting up and maintaining a Cloud project costs you in engineering time.

### Can I start on imaginary.app and move to Vertex AI later?

Yes. Both are async: submit, poll, download. Keep your prompt and settings in your own data model, and moving later means swapping the HTTP client and authentication, not your product logic.

How we checked: Google prices and model status are from the sources below, checked on November 8, 2026. The Google request shape is the one our backend sends to Veo 3.1. Our prices come from our pricing catalog. The setup-time example is an illustration, not a measurement.

## Sources

Prices, model specs and dates on this page were checked on November 8, 2026. Third-party prices change; check the source before you rely on a number.

- [Google Cloud — generative AI pricing](https://cloud.google.com/vertex-ai/generative-ai/pricing)
- [Google Cloud — Veo 3.1 model documentation](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/veo/3-1-generate)
- [Google Cloud — model versions and lifecycle](https://docs.cloud.google.com/gemini-enterprise-agent-platform/models/model-versions)
- [Google AI for Developers — Gemini API deprecations](https://aistudio.google.com/docs/deprecations)
- [imaginary.app pricing](https://www.imaginary.app/pricing)

Veo 3.1, no Cloud project

## Try the key-based route first

Packs start at $9.99 with no subscription, and failed generations are refunded. Move to Google Cloud later if your volume justifies it.

[Get API Access](https://www.imaginary.app/api)[Read the Docs](https://www.imaginary.app/api-docs)

Sign in with Google to create an account; buy credits only when you need them.
