StackMap
Subscribe

LangGraph alternatives

Curated alternatives to LangGraph — and why you'd switch.

CrewAI

Orchestrate role-playing, autonomous AI agents that collaborate on tasks.

Why switchHigher-level, opinionated multi-agent API vs. LangGraph's low-level control. Trade control for speed.
Full comparison →
DSPy

Program — don't prompt — your language models. Compile declarative pipelines into optimized prompts.

Why switchA different bet: optimize prompts as code rather than orchestrate them.
Full comparison →
deer-workflow

Graph-engineering runtime where TypeScript owns control flow and replaceable coding-agent CLIs — Codex, Claude Code, Pi — do the semantic work inside each node. Pilot for DeerFlow 3.0.

Why switchBoth make the graph rather than the prompt the unit of control. LangGraph is a library whose nodes call models directly, with checkpointing and durable execution; in deer-workflow a node shells out to an entire coding-agent CLI session, so the runtime is orchestrating harnesses, not model calls.
Full comparison →
eko

Production JavaScript framework for agentic workflows: one sentence becomes a dependency-aware multi-agent plan that runs in browser, Node, or extension — with pause/resume and snapshot recovery.

Why switchSame job — durable multi-step agent workflows. LangGraph is the Python graph incumbent; Eko is the JS-native answer that also runs inside browsers and extensions.
Full comparison →
labs-OO-Agents

NVIDIA's object-oriented agent framework: an agent is a Python class — fields are state, docstrings are prompts, and a `...` method body becomes an LLM loop that acts by writing Python.

Why switchOpposite shapes for the same job. LangGraph makes the topology explicit — nodes, edges, checkpoints, durable execution; NOOA hides it in Python semantics, where methods are the steps and `self` is the state. Choose LangGraph when the graph is the artifact you reason about, NOOA when the class is.
Full comparison →
MetaGPT

The original "AI software company" multi-agent framework — role-assigned agents (PM, architect, engineer) turn a one-line requirement into PRD, design, and code.

Why switchTrade-off runs the other way: LangGraph is low-level graph control you assemble; MetaGPT is a pre-built company you configure. Outgrowing MetaGPT's opinions usually lands you here.
Full comparison →
swiftide

Rust framework for LLM apps: an agent harness, compile-time-typed task graphs, and streaming RAG pipelines — MCP toolboxes, human-in-the-loop approval, tracing with Langfuse support.

Why switchSame job — durable, graph-orchestrated LLM apps with human-in-the-loop — different type systems: LangGraph gives you the Python/JS ecosystem's breadth; Swiftide gives you compile-time-typed nodes and hand-offs in Rust.
Full comparison →
agentfield

Open-source control plane that runs AI agents as microservices: write plain Python/Go/TS functions, get REST endpoints with routing, queues, retries, memory and tracing — one laptop to 10k agents.

Why switchSame job — production multi-agent systems — opposite shape. LangGraph is an in-process library: model your workflow as a stateful graph with durable execution. agentfield is an out-of-process control plane: plain functions become REST microservices and the platform handles fan-out, queues and retries, explicitly rejecting graph wiring. Library and embedded → LangGraph; platform and service-oriented → agentfield.
Full comparison →
agents-cli

CLI + agent-skills layer that turns your coding assistant into a Google Cloud agent-lifecycle expert: scaffold ADK projects, run and evaluate them, then deploy and publish to Gemini Enterprise.

Why switchBoth are ways to build production agentic apps, but from opposite ends: LangGraph is an open, embeddable graph runtime that runs anywhere, while agents-cli is Google's CLI+skills around ADK that scaffolds, evaluates, and deploys agents specifically on Google Cloud / Gemini Enterprise. Choose by ecosystem and portability needs.
Full comparison →