We may earn a commission if you buy through links on this page. It never changes a score or a ranking. How we make money
Pricing is quote-based.
Vertex AI has no free plan, but Google Cloud offers $300 in trial credits for 90 days. The pay-as-you-go plan also includes monthly allowances for selected Gemini search and grounding queries, Vizier trials and Agent Runtime usage. After those credits and allowances, billing depends on consumed resources such as tokens, node-hours and queries. Add-ons have separate rates, so a project’s total depends on its models, workload and use of grounding, search and pipeline services.
01Every plan, line by line
| Plan | Billed annually | Billed monthly | Built for |
|---|---|---|---|
| Pay-as-you-go (Consumption-based) | — | Quote | Teams paying for AI and ML usage |
Add-ons and overages
- Grounding with your data: $2.50 per 1,000 prompts
- Grounding with Google Search or Maps: $14 per 1,000 queries after the monthly free allowance
- Web Grounding for Enterprise: $45 per 1,000 grounded prompts
- Vertex AI Search: $1.50–$4.00 per 1,000 queries
- Vertex AI Pipelines: $0.03 per pipeline run
02What the plan grid doesn't tell you
Usage beyond allowances
The monthly query, Vizier and Agent Runtime allowances are limited; usage is billed by consumed resources, including tokens, node-hours and queries.
Grounding charges
Grounding with Google Search or Maps costs $14 per 1,000 queries after the monthly free allowance; other grounding options have separate rates.
Search and pipeline use
Vertex AI Search costs $1.50–$4.00 per 1,000 queries. Vertex AI Pipelines costs $0.03 per pipeline run.
Commitment pricing
One-year and three-year commitment discounts are available, but fixed prices are not published; request a quote for project-specific estimates.
03Which plan should you buy?
| If you are… | Get | Why |
|---|---|---|
| Side project using trial credits | Pay-as-you-go (Consumption-based) | Use the $300 trial credit for 90 days, while tracking consumption because the credit is not a spending cap. |
| Startup in production | Pay-as-you-go (Consumption-based) | There is no flat monthly fee; model token, compute and query usage, including add-ons, before launch. |
| Team with steady workloads | Pay-as-you-go (Consumption-based) | One-year and three-year commitment discounts are available, but compare project-specific estimates because fixed prices are not published. |
04Is Vertex AI worth the money?
Vertex AI can be worth paying for when a team needs managed training, model access, deployment and MLOps within Google Cloud. The $300 trial credit and monthly usage allowances can help with evaluation, but they do not set a maximum bill. Production costs depend on resource consumption and add-ons such as grounding, search and pipeline runs. There is no self-hosted Vertex AI option to avoid cloud usage charges; teams should forecast actual workloads and set billing controls before moving beyond evaluation.
Vertex AI pricing FAQ
How much does Vertex AI cost?
Vertex AI has no flat monthly or annual subscription and no published fixed starting price. The pay-as-you-go plan bills by resources consumed, including tokens, node-hours and queries. Add-ons have separate rates, such as $0.03 per Vertex AI Pipeline run and $1.50–$4.00 per 1,000 Vertex AI Search queries.
Does Vertex AI offer free credits?
Google Cloud offers $300 in trial credits for 90 days. The pay-as-you-go plan also lists 5,000 search or grounding queries per month across Gemini models, up to 100 Vizier trials per month, and Agent Runtime allowances of 50 vCPU-hours and 100 GiB-hours per month. Credits are not a spending cap.
What does grounding cost on Vertex AI?
Grounding with your data costs $2.50 per 1,000 prompts. Grounding with Google Search or Maps costs $14 per 1,000 queries after the monthly free allowance. Web Grounding for Enterprise costs $45 per 1,000 grounded prompts.
Are Vertex AI commitment discounts available?
Google lists one-year and three-year commitment discounts, but fixed prices are not published. Vertex AI has no standard monthly or annual subscription price. Ask Google Cloud for an estimate based on the services, resources and workload you expect to use.