The short answer
Google announced Gemini API rates of US$0.40 per generated second for Veo 3 and US$0.15 per second for Veo 3 Fast on September 8, 2025. Those are official historical rates—not a verified current quote. For production, there’s a key lifecycle caveat: Google Cloud’s documentation lists the veo-3.0-generate-001 and veo-3.0-fast-generate-001 Vertex AI endpoints for migration to Veo 3.1 before June 30, 2026. That date has passed as of November 9, 2026. (developers.googleblog.com)
So, don’t budget a new production integration using the old Veo 3 prices without first confirming the exact model ID and rate for your intended API route. The sources available here establish the old Gemini API rates and the Vertex AI migration date, but not a current official price for a replacement model.
Veo 3 cost by API route and version
| Model and route | Published price | What the price establishes |
|---|---|---|
| Veo 3, Gemini API | US$0.40 per generated second | Google announced this rate on September 8, 2025. It is not confirmation that the old model ID remains available or that this is today’s applicable rate. (developers.googleblog.com) |
| Veo 3 Fast, Gemini API | US$0.15 per generated second | Google announced this rate on September 8, 2025. Treat it as historical unless Google confirms it for your current model and route. (developers.googleblog.com) |
| Veo 3 and Veo 3 Fast, Vertex AI | No current price established here | Google’s model documentation lists the old endpoints for migration before June 30, 2026; it does not establish a current price for using them after that date. (docs.cloud.google.com) |
| Veo 3.1 or another replacement model | No current price established here | Google’s documentation identifies Veo 3.1 model IDs and capabilities, but the material captured for this article does not establish a current official price for those models. (docs.cloud.google.com) |
The versions and routes are not interchangeable: an older Veo 3 rate does not automatically apply to Veo 3.1, Vertex AI, a consumer subscription, or a third-party platform. Google’s Vertex AI documentation identifies veo-3.1-generate-001 and veo-3.1-fast-generate-001 separately from the older Veo 3 endpoints. (docs.cloud.google.com)
Example Veo 3 API costs
Use this formula for a simple estimate:
Estimated cost = published price per second × total generated seconds
Using Google’s September 2025 announced Gemini API rates:
| Total generated duration | Veo 3 at US$0.40/sec | Veo 3 Fast at US$0.15/sec |
|---|---|---|
| 8 seconds | US$3.20 | US$1.20 |
| 10 seconds | US$4.00 | US$1.50 |
| 30 seconds | US$12.00 | US$4.50 |
These are arithmetic illustrations based on historical rates, not current provider quotes or guaranteed invoice totals. The 10- and 30-second rows mean total generated duration—for example, across multiple clips—not necessarily one clip. Google Cloud’s Veo 3 documentation lists video lengths of 4, 6, or 8 seconds per output. (developers.googleblog.com)
The calculation also assumes that each generated second is billed at the stated rate. The sources here do not establish current billing details such as rounding, minimum charges, or how failed generations and retries are treated. Confirm those rules before using the estimate as a production budget.
API charges are not subscription or platform-credit prices
A consumer plan, API usage, and a third-party platform may each use a different billing unit. A monthly subscription or a balance of Flow or platform credits does not, by itself, tell you the dollar cost of a production API call. The official Google AI plans page describes consumer video-generation benefits, but that is not an API rate card; the captured plan information does not supply a Veo API price. (developers.googleblog.com)
Likewise, a third-party platform’s credits-per-second figure is that platform’s unit—not a Google API price. To compare routes, first identify the exact model, provider, billing unit, and terms, then compare costs on the same basis.
What to verify before choosing Veo for production
Before approving an implementation or budget, confirm these details in the official documentation for the specific provider and model you plan to call:
- Model ID and lifecycle: The old Vertex AI IDs
veo-3.0-generate-001andveo-3.0-fast-generate-001are listed for migration before June 30, 2026. The Veo 3.1 IDs are different. (docs.cloud.google.com) - Current rate and billing unit: Confirm the price for the selected model and route; don’t carry forward a September 2025 rate without confirmation. (
developers.googleblog.com) - Output fit: Google’s Veo 3.1 documentation lists 4-, 6-, and 8-second video lengths for the documented model, with supported aspect ratios and resolutions specified separately. Check the requirements for your actual workload rather than assuming that older Veo 3 capabilities or limits carry over. (
docs.cloud.google.com) - Billing details that affect totals: Confirm duration charging, any minimum or rounding rules, and the treatment of failed generations and retries. Those details are not established by the historical rate announcement cited here.
Bottom line: the defensible historical answer is US$0.40 per second for Veo 3 and US$0.15 per second for Veo 3 Fast on the Gemini API, announced September 8, 2025. Because Google’s Vertex AI documentation lists the old model endpoints for migration before June 30, 2026, those figures should not be presented as current production prices without confirmation for the exact model and route. (developers.googleblog.com)
Sources
Veo 3 and Veo 3 Fast – new pricing, new configurations and better resolution
Veo 3 | Gemini Enterprise Agent Platform