# Gemini platform An open Google Cloud platform for building, deploying, governing, and optimizing enterprise AI agents. - Findings24: - Site: - Preços: Freemium - Categorias: Criador de agentes de IA, Infraestrutura para agentes de IA, Implantação de modelos de IA ## O que é Gemini platform? Gemini platform is an open, integrated Google Cloud platform for technical teams building enterprise AI agents and applications. It combines agent development, model selection, customization, deployment, and governance in one environment, with support for agents grounded in enterprise data and used across applications and workflows. Teams can design and test agents in Agent Studio, select models and assets through Model Garden, or use custom training with their preferred frameworks and code. The platform also covers machine learning operations, including evaluation, workflow orchestration, model management, feature reuse, and monitoring. It is suited to developers, data scientists, and ML engineers working on production AI systems, from multimodal data extraction and classification to multi-step workflow automation. ## O que Gemini platform pode fazer? ### Agent development in Agent Studio Design, test, and manage prompts using natural language, code, images, or video. Agent Studio provides access to generative AI models, including Gemini models, for evaluation, tuning, and deployment. ### Model and asset selection Model Garden lets teams discover, test, customize, and deploy Google models, third-party models such as Anthropic’s Claude model family, and selected open-source models and assets. ### Custom model training Developers can use preferred machine learning frameworks, write their own training code, and choose hyperparameter-tuning options when managed model workflows do not provide enough control. ### Integrated data and AI workspaces Colab Enterprise and Workbench notebooks are integrated with BigQuery, providing a shared environment across data and AI workloads. Training and Prediction support model development and production deployment. ### MLOps lifecycle tools Purpose-built tools support model evaluation, pipeline orchestration, model registration, feature storage and reuse, and monitoring for input skew and drift across predictive and generative AI projects. ### Agent governance and orchestration Gemini can be used to securely register, manage, and govern custom-built agents. Google Antigravity provides a centralized application for steering, customizing, and orchestrating multiple agents. ## Casos de uso ### Build production AI agents Developers can build, customize, and fine-tune agents with frameworks such as the Agent Development Kit, then deploy them in applications and workflows on a platform intended for production use. ### Extract and classify multimodal data Teams can use Gemini models and Agent Studio prompts to extract text from images, answer questions about uploaded images or videos, summarize content, and classify data. ### Evaluate and select models Data scientists and ML engineers can compare models with Model Evaluation, explore available models in Model Garden, and choose a model or customization path for a specific use case. ### Automate multi-agent workflows Organizations can use Google Antigravity to coordinate multiple agents on workflows such as a product launch, including code generation, on-brand asset creation, and customer email production. ### Operate ML projects at scale ML teams can standardize projects with pipelines, model registry, feature storage, and monitoring, while using integrated notebooks and infrastructure for training and serving. ## Perguntas frequentes ### Who is Gemini platform for? It is intended for technical teams, including developers, data scientists, and ML engineers, who build, deploy, manage, or optimize enterprise AI agents and machine learning models. ### What are the main ways to build or customize models? Teams can use Agent Studio with generative AI models, select and customize models through Model Garden, or use custom training with their own frameworks, code, and hyperparameter-tuning choices. ### Can Gemini platform support multimodal workflows? Yes. Agent Studio supports prompts using natural language, code, images, or video, and the source describes workflows for extracting text from images and generating answers about uploaded images or videos. ### Does the platform include MLOps tools? Yes. The platform includes Model Evaluation, Pipelines, Model Registry, Feature Store, and monitoring for input skew and drift, along with training and prediction capabilities. ### How is Gemini platform priced? Gemini platform follows Google Cloud’s usage-based pricing approach. Google Cloud says customers pay for the services they use; product costs vary by product and usage. New Google Cloud customers can receive up to $300 in credits to try Gemini platform and other Google Cloud products. ## Informações rápidas - **Provider**: Google Cloud - **Category**: AI agent and machine learning platform - **Primary users**: Developers, data scientists, and ML engineers - **Core workflow**: Build, customize, deploy, govern, and monitor AI agents and models - **Model access**: Google, third-party, and selected open-source models through Model Garden - **Pricing model**: Google Cloud usage-based pricing; new customers may receive up to $300 in credits