Loading tool details…
Free deprecated educational framework that teaches multi-agent coordination fundamentals through minimal Agent and handoff abstractions.
OpenAI's teaching-oriented framework for learning multi-agent systems, now superseded for new supported development by the production-oriented OpenAI Agents SDK.
OpenAI Swarm is a free, MIT-licensed OpenAI GitHub repository for learning multi-agent handoffs, best used as an educational reference and prototype sandbox rather than a supported production framework because OpenAI now points new agent builders to the OpenAI Agents SDK. The repository is publicly available at github.com/openai/swarm, carries an MIT license, and is presented as an experimental educational framework rather than a maintained product line with paid tiers. Its core API is intentionally compact: an Agent combines instructions with functions, the run-style client loop executes conversations, context variables are passed as dictionaries, and handoffs happen when a function returns another Agent. Those facts make Swarm useful for inspecting the mechanics of agent routing without the larger surface area of a production orchestration stack. Swarm also includes example-oriented patterns for tasks such as triage, support-style routing, and shopping or customer-service flows, giving learners concrete reference implementations instead of only conceptual documentation. The important 2026 status is that Swarm should be treated as a legacy learning resource in the OpenAI ecosystem: its design ideas remain useful, but OpenAI directs developers who are starting new supported agent projects toward the OpenAI Agents SDK. That distinction matters for accuracy. Swarm being MIT licensed means commercial use of the code is permitted under the license, but the framework itself should not be described as production-ready without additional engineering, security review, state management, observability, retries, evaluation, and operational guardrails. Use Swarm when you want a small, readable Python codebase for understanding explicit handoffs, tool-style functions, stateless orchestration, and context passing. Avoid it when your project needs officially supported production workflows, documented scalability guarantees, persistent state, enterprise support, native MCP integration, tracing, guardrails, or the richer agent primitives available in current maintained frameworks. In short, Swarm is valuable as a free historical and educational reference, but it is not the best default choice for new production OpenAI agent applications.
An educational framework for understanding multi-agent orchestration patterns through minimal, readable Python abstractions. Use it for learning and historical comparison, not for new production systems.
Uses an Agent class pattern combining instructions with executable functions, making agent behavior easy to inspect and teach.
Uses explicit handoff functions for agent-to-agent task transfer, creating a readable coordination pattern for examples and prototypes.
Demonstrates explicit state management through intentionally stateless design, helping learners see what production frameworks must add.
Implements simple dictionary-based context passing that teaches fundamental inter-agent communication patterns.
Provides educational examples covering customer service, shopping assistance, and support-style coordination scenarios.
Shows relatively direct OpenAI Chat Completions API usage, helping developers understand the underlying calls behind agent behavior.
Free
Ready to get started with OpenAI Swarm?
View Pricing Options →OpenAI Swarm works with these platforms and services:
We believe in transparent reviews. Here's what OpenAI Swarm doesn't handle well:
In 2026, the important status note is that OpenAI Swarm should be treated as a deprecated educational repository. Developers starting new agent projects should evaluate the OpenAI Agents SDK or another maintained production framework.
AI Agent Builders
OpenAI Agents SDK is an open-source Python framework for building agentic apps with handoffs, guardrails, sessions, tracing, MCP tools, sandbox agents, and realtime voice agents.
AI agent framework
LangGraph is LangChain’s framework for reliable agents with low-level control, deployment, observability, evaluation, sandboxes and enterprise LangSmith services.
AI Development
Leading multi-agent framework and platform for building, deploying, and monitoring collaborative AI agent teams.
Multi-Agent Builders
Microsoft's open-source framework for building multi-agent AI systems with asynchronous, event-driven architecture.
Multi-Agent Builders
Agency Swarm is an open-source Python framework for building collaborative agent teams ("agencies") on top of the OpenAI Agents SDK / Responses API, with role-based agents, Pydantic-typed tools, and production-focused deployment patterns.
Pricing and features change often. If anything on this OpenAI Swarm page is wrong or missing, tell us and include a link to the source. We review every correction.