Vertex AI: Google Cloud's AI toolbox now lives inside Gemini Enterprise Agent Platform
Updated on September 12, 2026
Vertex AI is Google Cloud's artificial intelligence platform for building, deploying and monitoring machine learning models and agents in production. Google has folded it into the Gemini Enterprise Agent Platform, its umbrella brand for enterprise AI, and every familiar building block survived the move. Billing is pay-as-you-go, with $300 in trial credits for new Google Cloud accounts. Behind one console sit over 200 models, from Gemini to Claude and Llama.
- Over 200 models available through Model Garden
- EU data residency, including a Paris region
- Enterprise-grade IAM, VPC controls and SLAs
- Prompts never reused to train foundation models
- Billing spread across many meters, worth watching closely
- Steep learning curve, aimed at technical teams
- Deployed endpoints keep billing while idle
Inside Vertex AI, from Model Garden to Agent Builder
Model Garden lists more than 200 models, Google's own Gemini, Imagen and Veo alongside third-party options such as Anthropic's Claude, Meta's Llama, Mistral and the open Gemma family.
You paste a prompt into Vertex AI Studio, swap models from a dropdown, then push the winning setup to an API endpoint. From there, the rest of the toolbox takes over.
- Vertex AI Studio for prototyping text, image and code prompts
- Agent Builder for agents grounded in your own data
- Vertex AI Search for querying internal documents
- AutoML and pipelines for custom model training
From Vertex AI to the Gemini Enterprise Agent Platform
Vertex AI's standalone roadmap has ended, its services reorganized under the Gemini Enterprise Agent Platform unveiled at Google's Cloud Next conference. Nothing was retired in the process, and the APIs kept answering at the same addresses (do check your Python imports though, some older SDK modules carry a migration deadline).
Amazon answers with Bedrock, Microsoft with Azure AI Foundry, and that three-way race now shapes a large share of the tools built for developers. Google keeps a practical edge for European teams, with EU regions such as Paris and Belgium for data residency.
Metered billing by token, node-hour and query
Vertex AI has no flat fee, every service runs its own meter. Models bill per million tokens, with Gemini 3.1 Pro at $2 in and $12 out, Gemini 3.8 Flash at an introductory $0.75 and $3.75, and Claude Sonnet 4.6 at $3 and $15. Watch the meters, a forgotten endpoint keeps charging even when nobody calls it.
For a quick prototype, Google AI Studio remains the free playground, and Vertex AI takes over once you want SLAs, VPC controls and fine-grained access rights. Rates shift with every model release, so treat Google Cloud's published price list as the only figure worth budgeting on.
| Service | Billing unit | Ballpark |
|---|---|---|
| Generative models | per million tokens | from $0.75 input |
| Vertex AI Search | per query | about $4 per 1,000 |
| Training and endpoints | per node-hour | T4 GPU around $0.35/hour |
| New customer credits | one-off | $300 valid 90 days |
Frequently asked questions
What is Vertex AI?
Vertex AI is Google Cloud's platform for building, deploying and governing AI models and agents in production. Launched in 2021, it bundles access to Google's Gemini, Imagen and Veo models plus third-party ones like Claude and Llama, and now operates inside the Gemini Enterprise Agent Platform.
Is Vertex AI free?
No, billing is usage-based, though several free doors exist. New Google Cloud accounts get $300 in credits valid for 90 days, an Express Mode lets you try Studio and Agent Builder with limited quotas before enabling billing, and Vertex AI Search includes a batch of free queries.
What is the difference between Vertex AI and Google AI Studio?
Google AI Studio is the free sandbox for testing Gemini models with a simple API key, while Vertex AI targets production with service accounts, IAM, VPC and SLAs. One detail worth knowing, free-tier AI Studio prompts may be used to improve Google products, whereas prompts sent through Vertex AI never are.
Should I use Vertex AI or Gemini?
Gemini is the consumer app with monthly subscriptions, while Vertex AI is the cloud platform exposing the same models through APIs for your own software. Chatting with an assistant calls for the Gemini app, shipping those models inside a product with enterprise guarantees calls for Vertex AI.
Verdict: Production is where this platform earns its keep. Engineering teams already on Google Cloud, and anyone who wants Gemini, Claude and Llama behind a single enterprise contract, will feel at home here.
