Google ADK vs LangGraph: graph workflows from two directions
Google's Agent Development Kit is an Apache-2.0 framework in Python, TypeScript, Go, Java and Kotlin with graph workflows since 2.0 and full integration with Google's Agent Runtime, while LangGraph is an MIT-licensed Python and TypeScript runtime that runs anywhere, so ADK fits teams on Google Cloud or outside Python and LangGraph fits teams that want a cloud-neutral graph runtime.
Checked
ADK 2.0 brought graph-based workflows, which puts it in direct overlap with LangGraph. Both now let a team combine deterministic code with model decisions in one graph, and both are model-agnostic in principle.
The differences are language reach, where each one is easiest to deploy, and maturity of the current major version. Google recommends ADK for agent development on Google Cloud, and Google's Agent Runtime also hosts LangGraph through managed templates, so the cloud does not force the choice.
What each offers
Google ADK
Graph workflows (new in 2.0), single and multi-agent orchestration, sessions, state and memory, context compression and caching, evaluation with user and environment simulation, A2A support, and a web dev UI. Model adapters cover Gemini, Gemma, Claude, OpenAI, Ollama, vLLM and LiteLLM.
LangGraph
Durable execution, persistence, human-in-the-loop editing of state, short-term and long-term memory, streaming, and debugging through LangSmith. LangChain's create_agent runs on it.
Deployment and maturity
ADK deploys to Agent Runtime on Gemini Enterprise Agent Platform, Cloud Run or GKE, or to a team's own infrastructure, and has full integration on Agent Runtime. LangGraph runs anywhere Python or Node runs, and Agent Runtime supports it through managed SDK templates.
ADK 2.0 reached general availability for Python on 19 May 2026, Go on 30 June 2026 and TypeScript on 21 August 2026, with breaking changes from 1.x. LangGraph has been at 1.x since 17 October 2025 with core graph APIs unchanged.
Which fits which job
ADK fits
Teams on Google Cloud, teams writing agents in Go, Java or Kotlin, and designs where agents talk to other agents over A2A.
LangGraph fits
Teams that want a cloud-neutral runtime with a long track record of durable, human-in-the-loop workflows and the LangChain integration surface.
How Motiontic approaches it
Motiontic's engineers are framework-neutral. They choose per project after a short spike on the client's real task, keep prompts, tool definitions and evals outside the framework so a later switch is a refactor rather than a rewrite, and hand the eval suite to the client's team so the next framework decision can be measured.
Google ADK and LangGraph, figures retrieved 30 October 2026
| Google ADK | LangGraph | |
|---|---|---|
| Licence | Apache-2.0 | MIT |
| Languages | Python, TypeScript, Go, Java, Kotlin | Python, JavaScript / TypeScript |
| Graph workflows | Yes, new in ADK 2.0 | Core model since launch |
| Latest Python version | 2.13.0 (23 Oct 2026); 2.0 GA 19 May 2026 | 1.2.18 (26 Oct 2026); 1.0 on 17 Oct 2025 |
| State and durability | Sessions, state, memory, context compression | Durable execution, persistence, resume after failure |
| Evaluation and tooling | Built-in evaluation, web dev UI | LangSmith (commercial) |
| Google Agent Runtime | Full integration | Managed SDK template |
| GitHub stars (Python repo) | 22,050 | 44,172 |
| PyPI downloads, last week | 2.45M | 13.76M |
Common questions
Does Google ADK only work with Gemini?
No. ADK's docs list adapters for Gemma, Claude, OpenAI, Ollama, vLLM, LiteLLM and LiteRT-LM alongside Gemini.
Can LangGraph agents run on Google Cloud?
Yes. Google's Agent Runtime supports LangGraph through managed SDK templates, alongside LangChain, AG2 and LlamaIndex. ADK has full integration.
Does Google ADK have graph workflows like LangGraph?
Yes. Graph-based workflows that combine deterministic code with model reasoning arrived in ADK 2.0, which reached general availability for Python on 19 May 2026.
Which languages does Google ADK support?
Python, TypeScript, Go, Java and Kotlin, which is wider language coverage than LangGraph's Python and TypeScript.
Related
AI agent development
Agents that do real work for users, with the evals, traces and guardrails to keep them in production.
LangChain vs LangGraph
The agent layer and the runtime underneath it, compared after the 1.0 releases.
MCP vs A2A
The protocol for tools and data, compared with the protocol for agents talking to agents.
A2A protocol
The open protocol for agents talking to other agents, now under the Linux Foundation.
Agent loop
Model call, tool call, result, repeat: the cycle at the centre of every agent.
Running agents that have to hold up?
Describe the agent, the systems it touches and what it must never get wrong. The first call is with a lab engineer, who will say plainly whether it is ready for production, and whether Motiontic is the right team.