The Swarms API Course
A four-part curriculum that takes you from your first API call to production multi-agent systems. Four parts, 31 lessons, 12 live API trials you can run directly in the page, and a checkpoint project plus graded quiz at the end of each part. Earn points as you go and rank up from Recruit to Swarm Architect.
Your progress
Points and lesson progress are tracked in your browser: no account needed. Rank up from Recruit to Swarm Architect by completing lessons, running live trials, and acing the quizzes. Finish all lessons, the checkpoint, and the quiz in a part to earn its badge.
Curriculum
From first request to production swarms.
Each part builds on the last. Work through them in order, run the live trials as you go, and finish each checkpoint project and quiz before moving on.
Foundations: Your First Agent
Understand what the Swarms API is, get authenticated, and run a single AI agent end to end. By the end of this part you will have executed your first agent completion and know how to read every field in the response.
- What is the Swarms API?
- Get your API key
- Your first request: the health check
- Anatomy of an agent
- Run your first agent completion
- Discover models and check your balance
- Reading responses and handling failures
Single-Agent Capabilities
One agent can do far more than answer a prompt. This part upgrades your agent with conversation memory, real-time streaming, schema-enforced JSON, function tools, live web search, MCP connections, and fully autonomous multi-step execution.
- Multi-turn conversations
- Streaming responses
- Structured outputs
- Web search and vision
- MCP: connect agents to your systems
- Autonomous agents and sub-agent delegation
- The OpenAI-compatible endpoint
- Reusing configurations and marketplace prompts
Multi-Agent Orchestration
Complex work benefits from specialists. This part introduces the swarm completions endpoint and walks through the major orchestration architectures: sequential pipelines, parallel fan-outs, hierarchical teams, collaborative patterns, and directed graphs. You will learn to pick the right one on sight.
- From one agent to a swarm
- SequentialWorkflow: the pipeline
- ConcurrentWorkflow: the fan-out
- HierarchicalSwarm: the team with a director
- Collaboration patterns: debate, consensus, and councils
- Routing and dynamic architectures
- GraphWorkflow: directed acyclic graphs
- Streaming swarms and running multiple tasks
- Choosing the right architecture
Production: Scale, Cost, and Operations
The gap between a working demo and a production system is operational: processing thousands of tasks, surviving rate limits, controlling spend, and knowing what your agents did last night. This part covers the endpoints and habits that close the gap.
- Batch processing at scale
- Reasoning agents
- Rate limits, retries, and error handling
- Cost management
- Observability: logs and metrics
- Regions, priority, and security
- The production readiness checklist
FAQ
About the Swarms API course.
How long does the Swarms API course take?
About six hours across four parts: Foundations (60 minutes), Single-Agent Capabilities (90 minutes), Multi-Agent Orchestration (105 minutes), and Production (90 minutes). Each part ends with a checkpoint project and a graded quiz.
Is the Swarms API course free?
Yes, the full course is free with no signup. The embedded live API trials run against your own Swarms API key; new accounts include free credits, and every trial is labeled with whether it consumes credits.
What do I need before starting?
Basic Python and a free Swarms API key from swarms.world/platform/api-keys. Every example also translates directly to any HTTP client or the OpenAI SDK, so Python is a convenience rather than a requirement.
Does the course cover multi-agent orchestration?
Yes. Part 3 is a full multi-agent orchestration module covering SequentialWorkflow, ConcurrentWorkflow, HierarchicalSwarm, GroupChat, MixtureOfAgents, MajorityVoting, debate and council patterns, AgentRearrange, and GraphWorkflow DAGs, with guidance on choosing the right architecture.
Can I run real API requests inside the course?
Yes. The course embeds twelve live API trials: paste your API key once and execute real requests against api.swarms.world directly from the lesson, edit the JSON request bodies, and inspect status codes, latency, and responses inline.
Does the course track my progress?
Yes. The course tracks completed lessons, checkpoint projects, live trial runs, and quiz scores, and awards points for each: 10 per lesson, 15 per successful API trial, 25 per checkpoint, and 5 per correct quiz answer with a perfect-score bonus. Points rank you from Recruit up to Swarm Architect, and finishing every activity in a part earns its badge. Progress is stored in your browser, no account required.
What should I do after finishing the course?
Ship something. The production part ends with an overnight batch pipeline capstone, and the Swarms API documentation at docs.swarms.ai continues with an examples library covering finance, healthcare, legal, and operations swarms, plus migration guides from LangGraph, CrewAI, and AutoGen.
Before you start
- Basic Python (every example also translates directly to any HTTP client)
- A free Swarms API key from swarms.world/platform/api-keys
- No prior experience with AI agents required
Ready when you are
Part 1 takes about an hour and ends with a working agent you built and ran yourself.