Google's open-source framework for building, evaluating, and deploying multi-agent AI systems with Gemini and other LLMs.
Google's toolkit for building AI agents that can use Google's AI models, tools, and services.
Google Agent Development Kit (ADK) is a free, open-source multi-agent AI framework in the AI Agent Builders category, licensed under Apache 2.0, that provides Python, TypeScript, Go, and Java SDKs for building, evaluating, and deploying production-grade agent systems powered by Gemini and other LLMs at no cost for the framework itself.
ADK offers structured agent development with built-in evaluation, debugging tools, and deployment options that set it apart from more general-purpose orchestration frameworks. The framework ships with a comprehensive evaluation system featuring trajectory accuracy scoring, user simulation, environment simulation, and custom metrics — capabilities that are rare among the 30+ agent builders in the market today.
At its core, ADK provides multiple agent types including LLM agents, sequential workflow agents, parallel execution agents, loop agents, and custom agents. These can be composed into hierarchical multi-agent teams where agents delegate tasks to sub-agents and coordinate complex workflows. The Python 2.0 Beta introduces new workflow primitives and agent teams functionality, while TypeScript 1.0 brings first-party JavaScript and Node.js support.
Developers get three runtime modes out of the box: a web-based debugging UI for visual inspection of agent interactions and tool calls, a CLI for terminal-driven workflows, and an API server for production integrations. The local web UI is particularly valuable for understanding agent decision-making and debugging multi-agent coordination in real time.
ADK natively supports the Model Context Protocol (MCP), enabling agents to consume any MCP-compatible tool server without custom integration code. Combined with OpenAPI tool generation and traditional function tools with action confirmations, ADK provides flexible tooling options. Model support extends beyond Gemini through LiteLLM integration to include Claude, GPT-4, Ollama, vLLM, and other providers.
For production deployment, teams can choose between Google Cloud's Vertex AI Agent Engine for managed scaling with enterprise security, IAM, and monitoring, or self-host on Cloud Run, GKE, Docker, or any container platform. Observability is built in with logging, metrics, and OpenTelemetry trace support across all deployment modes.
Was this helpful?
**Best multi-agent framework for Google Cloud ecosystems.** Google ADK's built-in evaluation tools, local debugging UI, and native MCP support provide a more structured development experience than LangChain or CrewAI. With four official SDKs (Python, TypeScript, Go, Java), it is one of the most polyglot agent frameworks available. The primary trade-off is a smaller third-party ecosystem due to the framework's youth (launched April 2025) and the strongest feature integration being tied to Gemini and Google Cloud services. Teams already invested in Google Cloud will find ADK's Vertex AI Agent Engine deployment path and native Workspace integrations compelling, while teams on other clouds should weigh the Gemini-optimized features they may not fully leverage.
Build hierarchical agent teams using sequential, parallel, and loop workflow agents alongside LLM agents and custom agents. Agents can delegate to sub-agents through agent routing, and the new Python 2.0 Beta introduces agent teams for coordinated multi-agent execution with shared state and task delegation patterns.
Test agent performance with criteria-based scoring, user simulation, environment simulation, custom metrics, and an optimization module for iterative improvement. Score trajectory accuracy (correct steps taken), tool usage quality, and final response relevance — capabilities that most competing frameworks require third-party integrations to achieve.
ADK natively supports the Model Context Protocol, allowing agents to consume any MCP-compatible tool server without custom integration code. Combined with OpenAPI tool generation and traditional function tools with action confirmations, ADK provides one of the most flexible tooling ecosystems among agent frameworks.
Ships with three runtime modes: a web-based UI for visual debugging of agent interactions and tool calls, a CLI for terminal-driven workflows, and an API server for production integrations. The web UI provides real-time inspection of agent decision-making, making it significantly easier to debug multi-agent coordination than log-based approaches.
Available in four official SDKs — Python 2.0 Beta, TypeScript 1.0 (new in 2026), Go, and Java — all maintained under github.com/google. Supports Gemini, Gemma, Claude, Ollama, vLLM, LiteLLM, LiteRT-LM, and Apigee AI Gateway for flexible model selection across providers.
$0
Pay-as-you-go (Google Cloud usage)
Ready to get started with Google Agent Development Kit (ADK)?
View Pricing Options →We believe in transparent reviews. Here's what Google Agent Development Kit (ADK) doesn't handle well:
Weekly insights on the latest AI tools, features, and trends delivered to your inbox.
ADK TypeScript 1.0 is now available as of 2026, opening the framework to JavaScript/Node.js teams alongside the existing Python, Go, and Java SDKs. ADK Python 2.0 Beta also launched with new workflow primitives, agent teams, ambient agents, resumable runs, and cancelable execution — significantly expanding the framework's orchestration capabilities.
AI Agent Builders
Open-source Python framework that orchestrates autonomous AI agents collaborating as teams to accomplish complex workflows. Define agents with specific roles and goals, then organize them into crews that execute sequential or parallel tasks. Agents delegate work, share context, and complete multi-step processes like market research, content creation, and data analysis. Supports 100+ LLM providers through LiteLLM integration and includes memory systems for agent learning. Features 48K+ GitHub stars with active community.
AI Agent Builders
Graph-based workflow orchestration framework for building reliable, production-ready AI agents with deterministic state machines, human-in-the-loop capabilities, and comprehensive observability through LangSmith integration.
Multi-Agent Builders
Microsoft's open-source framework for building multi-agent AI systems with asynchronous, event-driven architecture.
AI Agent Builders
AWS open-source SDK for building AI agents in Python and TypeScript with model-driven tool orchestration, multi-provider LLM support, and native AWS deployment options.
AI Agent Builders
OpenAI's official open-source framework for building agentic AI applications with minimal abstractions. Production-ready successor to Swarm, providing agents, handoffs, guardrails, and tracing primitives that work with Python and TypeScript.
No reviews yet. Be the first to share your experience!
Get started with Google Agent Development Kit (ADK) and see if it's the right fit for your needs.
Get Started →Take our 60-second quiz to get personalized tool recommendations
Find Your Perfect AI Stack →Explore 20 ready-to-deploy AI agent templates for sales, support, dev, research, and operations.
Browse Agent Templates →