Vertex AI Studio

Vertex AI Studio is a Google Cloud console tool for rapidly prototyping and testing generative AI models. It helps you streamline foundation model workflows by letting you test, tune, and deploy enterprise-ready generative AI. Google’s Vertex AI Studio offers an efficient experience for discovering and refining models, settings, system instructions, and prompts through built-in experiences and a dedicated collaborative workspace environment.

Vertex AI Key Features

• Gemini multimodal models – Access to the latest Gemini models from Google, including Gemini 3. Prompt and test Gemini using text, images, video, or code. Try sample prompts for extracting text from images, converting image text to JSON, and generating answers about uploaded images.

• Access to 200+ models – Choose from proprietary models, open models, and third-party models through Vertex AI’s Model Garden. Includes Google’s foundation models as APIs and Gemma, a family of lightweight open models. Vertex AI Studio supports certain third-party models offered as models as a service (MaaS), such as Anthropic Claude models and Meta’s Llama models.

• Prompt design – Adapt models to your use case through a familiar chat interface. Adjust response parameters like temperature to elicit more creative responses.

• Model tuning – Improve response quality by tuning foundation models with your own data. Access tuning options like adapter tuning, Reinforcement Learning from Human Feedback (RLHF), and style and subject tuning for image generation.

• Vertex AI Extensions – Connect models to proprietary data sources or third-party services. Create applications that deliver real-time information, incorporate company data, and take action on the user’s behalf.

• MLOps integration – Deploy generative capabilities with a few lines of code. Scale and manage models in production using Vertex AI’s end-to-end MLOps capabilities and fully managed AI infrastructure.

• Enterprise-grade data governance – Data is protected, secure, and private when customizing models. None of the customer’s data, model weights, or input prompts are used to tune the original foundation models. When enterprises tune a model with their own data, the original model remains unchanged, and the new model never leaves your company’s environment.

• Grounding – Ground responses using Google Search, Maps, or your own data.

• Thinking capabilities – Solve complex requests and show the model’s thought process behind its response using Gemini’s built-in reasoning capabilities.

• Live API – Provide end users with the experience of natural, human-like voice conversations.



• Image and video generation – Use Imagen for image generation and Veo for video generation from text or image prompts.

• Generative AI evaluation service – Evaluate any generative model or application and benchmark results against your own judgment using your own evaluation criteria.

Vertex AI Roles

Onboarding and Administrative Features

Studio Discovery and Development Features

Built-in Tools and Capabilities

Vertex AI Common Uses

• Build with Gemini – Use Vertex AI Studio to access Gemini’s multimodal models. Prompt and test using natural language, text, code, or an image. Test, tune, and deploy enterprise-ready models.

• Design and test prompts – Learn basic concepts for designing well-structured prompts. Generate text, embeddings, code, images, videos, music, and more. Use the AI-powered prompting tool, browse the prompt gallery, and review prompting strategies.

• Tune foundation models from your data – Fine-tune large language models with your own data to achieve superior performance on specific tasks. Benefits include more accurate results, increased model robustness, reduced latency, and lower cost due to shorter prompts. Resources include tuning Gemini models, text tuning, and image tuning.

• Evaluate and optimize performance – Use the Gen AI evaluation service to evaluate any generative model or application. View and interpret evaluation results, prepare your evaluation dataset, and run evaluations.

• Generate a solution – Describe your problem (e.g., “I want to build a web app”) and receive a step-by-step guide, reference architecture, and available pre-built solutions. Note: You must be 18 or older to use this. Do not enter sensitive, confidential, or personal information.

Getting Started in Vertex AI

Prerequisites

1. Create a Google Cloud account (get $300 in free credits)

2. Select or create a Google Cloud project

3. Enable billing

4. Enable the Vertex AI API

5. Required IAM roles:

   • serviceusage.serviceUsageAdmin (to enable the Vertex AI API if not already enabled)

   • roles/aiplatform.user (to run prompts in Vertex AI Studio)

Test Gemini Using Sample Prompts

Summarization text prompt:

1. Go to the Prompt gallery page from the Vertex AI section in the Google Cloud console

2. In the Tasks drop-down menu, select Summarize

3. Open the Audio summarization card

4. Review the prompt and audio file (model is set to Gemini-2.0-flash-001 by default)

5. Click Submit to generate a summary in a bulleted list

6. To view the API code, click Build with code > Get code and select your preferred language

Code generation prompt:

1. Go to the Prompt gallery page from the Vertex AI section in the Google Cloud console



2. In the Tasks drop-down menu, select Code

3. Open the Generate code from comments card

4. Review the system instruction and incomplete Java methods with <WRITE CODE HERE> placeholders

5. Click Submit to generate the missing code

6. To view the API code, click Build with code > Get code and select your preferred language

References

Vertex AI Studio product page

Vertex AI Studio capabilities

Generative AI on Vertex AI overview

Quickstart: Send text prompts to Gemini using Vertex AI Studio