All articlesArticle

Google ADK vs Claude Agent SDK (2026)

· 6 min read · Matteo Gazzurelli

Two glass pavilions side by side on a wet square at night: one lit, with rows of workbenches inside, the other empty and dark

Google ADK and the Claude Agent SDK solve different problems, and the comparison only holds when you see which one. ADK is a framework for building agents and the workflows around them, in five languages, on any model, with a managed runtime on Google Cloud. The Claude Agent SDK is the agent loop of Claude Code as a library: a ready-made agent with file, shell and web tools, which you steer and embed. I have shipped work on both. Today my production work runs on Google ADK, and this article explains the comparison and the reasons.

As of November 2026. Both projects release often; the documentation linked below is the source of truth.

The short answer

  • Choose Google ADK when you are building an agent into a business process: several steps, some fixed and some that need judgement, tools that act on company systems, a team that has to maintain it for years, and a cloud where identity and logging already live.
  • Choose the Claude Agent SDK when the agent's job looks like Claude Code's job: reading and editing files, running commands, researching, on a Claude model, and you want that loop ready on day one instead of writing it.

Side by side

Google ADKClaude Agent SDK
What it isOpen-source framework to define agents, tools and workflowsThe Claude Code agent loop, tools and context management, as a library
LanguagesPython, TypeScript, Go, Java, KotlinPython, TypeScript (other languages through the CLI in headless mode)
ModelsGemini by default; also Claude, Gemma, OpenAI, Ollama, vLLM and others through LiteLLMClaude, through the Anthropic API, Amazon Bedrock, Google Cloud, Microsoft Foundry or Claude Platform on AWS
OrchestrationLLM agents plus workflow agents (sequential, parallel, loop); ADK 2.0 adds graph workflows that mix deterministic code and model stepsThe model plans its own steps; subagents for focused subtasks
Built-in toolsGoogle Search and other Google tools; your functions become tools from their signature and docstringRead, write and edit files, run commands, search the web
Control pointsCallbacks before and after each agent, model and tool callPermission modes, allow and deny rules, hooks around each tool call
External systemsMCP tools; A2A for agent-to-agent callsMCP servers
HostingAnywhere a container runs; on Google Cloud: Agent Runtime, Cloud Run, GKEWherever you run the process; Claude Managed Agents if you want Anthropic to host the loop
Managed servicesSessions, Memory Bank, Agent Identity and traces on Gemini Enterprise Agent PlatformSessions you can resume or fork; skills, commands and memory loaded from .claude/

The same agent, minimal code

Google ADK (Python)

pip install google-adk
adk create my_agent
# my_agent/agent.py
from google.adk.agents.llm_agent import Agent

def get_order_status(order_id: str) -> dict:
    """Returns the status of an order."""
    return {"order_id": order_id, "status": "shipped"}

root_agent = Agent(
    model="gemini-flash-latest",
    name="orders_agent",
    description="Answers questions about orders.",
    instruction="Answer questions about orders. Use get_order_status to look an order up.",
    tools=[get_order_status],
)

adk run my_agent runs it in the terminal, adk web opens a local debugging UI. The function is the tool: ADK reads its name, type hints and docstring. From here you add workflow agents around it, callbacks for permissions and approvals, and deploy the same code to Cloud Run or Agent Runtime.

Claude Agent SDK (Python)

pip install claude-agent-sdk
export ANTHROPIC_API_KEY=...
import asyncio
from claude_agent_sdk import query, ClaudeAgentOptions, ResultMessage

async def main():
    async for message in query(
        prompt="Review utils.py for bugs that would cause crashes. Fix what you find.",
        options=ClaudeAgentOptions(
            allowed_tools=["Read", "Edit", "Glob"],
            permission_mode="acceptEdits",
        ),
    ):
        if isinstance(message, ResultMessage):
            print(message.subtype)

asyncio.run(main())

The two examples show the difference better than any table. In ADK you describe the agent and hand it your tools. In the Agent SDK the agent already exists, with its tools, and you decide what it may touch.

Why my production work moved to ADK

Until this autumn I ran production work on both, one per vendor, as I wrote in From CrewAI to production agents. Four things moved the balance.

  1. The processes I automate are business processes. Most steps are fixed and an auditor expects them in the same order every time. ADK's workflow agents, and now graph workflows, keep those steps in code and give the model only the steps that need judgement. In the Agent SDK the model plans the steps, which is the right design for a coding agent and the harder one to certify for an order-to-invoice process.
  2. Model choice stays open. ADK runs Gemini by default and Claude when Claude tests best on a step. The Agent SDK is built around Claude. With ADK the model is a setting I change after an evaluation, and the framework stays.
  3. The production floor comes with it. On Gemini Enterprise Agent Platform the same ADK agent gets managed sessions, Memory Bank, its own identity and traces. With the Agent SDK I assemble those pieces myself, or move to Claude Managed Agents, which is a different service.
  4. Demand. The companies I talk to, in Italy and abroad, ask about Gemini, Google Workspace and Google Cloud. Requests for agents built on the Claude Agent SDK have been rare among my clients and contacts. A framework a client's team can find people for, and that their cloud vendor supports, is easier to hand over.

The Claude Agent SDK stays the tool I reach for when the job is the one it was built for: an agent that works on a codebase or a set of documents with file and shell tools. And Claude stays in my model shortlist, inside ADK.

Mixing them

The choice is less binary than the title suggests. Tools exposed through MCP serve both frameworks, so an integration written once survives a change of framework. ADK agents can call agents built elsewhere through A2A. A common shape in my projects: ADK for orchestration, identity and the fixed steps, with a Claude model on the step where it tests best.

I compare the full stacks, managed platform included, in the Google vs Anthropic guide. The Google ADK article maps where ADK sits inside Gemini Enterprise Agent Platform, and the Google ADK integration page describes how I take an ADK agent from demo to production.


Last technically reviewed: 3 November 2026. Sources: adk.dev, ADK Python quickstart, Claude Agent SDK overview, Agent SDK quickstart.

Written by

Matteo Gazzurelli, fractional CTO and agentic AI consultant with MaGa Srl, Brescia.

More about me

Agents in the Room

Pieces like this one, by email: an idea, a lesson from a project, a signal from the stack.