APIs and reference

Gemini Enterprise Agent Platform provides model APIs, platform REST and RPC APIs, client libraries, and command-line tools to build, deploy, and manage AI agents, generative AI applications, and machine learning workflows. This page highlights key model API references—such as the Interactions API, Live API, and APIs for embeddings, tuning, and evaluation—alongside RAG and platform SDK references across Agent Platform.

Model APIs

Interact with Gemini and other generative AI models for text and multimodal generation, real-time streaming, embeddings, customization, and evaluation.

Use a unified, stateful, and streaming interface to interact with models and agents.

Stream low-latency, bidirectional voice and video interactions with Gemini models.

Generate text and multimodal embeddings for semantic search, classification, and retrieval.

Customize model behavior using supervised fine-tuning.

Evaluate generative AI models and agents programmatically using rubrics and metrics.

RAG APIs

Ground model responses and connect your agents to external data sources using Retrieval-Augmented Generation (RAG).

Manage RAG corpora, import files, and retrieve relevant context with RAG Engine.

Understand the structure of retrieval queries and grounded generation responses.

Platform APIs

Use SDKs, command-line tools, and service-level APIs to manage Agent Platform resources, infrastructure, and developer environments.

Build with the Google Gen AI SDK and Agent Platform SDK in Python, Go, Java, Node.js, and C#.

Explore all HTTP REST resources and methods for Agent Platform services.

View gRPC service definitions and protocol buffer messages for Agent Platform.

Manage Agent Platform resources from the command line.

Prototype and test generative AI capabilities using streamlined REST endpoints and an API key.

Connect Model Context Protocol (MCP) clients and agents to Agent Platform tools.

Build, manage, and query high-scale vector similarity search indexes.

Configure and govern network traffic between agents, MCP servers, and external tools.

Manage notebook instances and development environments using REST, RPC, CLI, and client libraries.