OpenAI Agents SDK
The OpenAI Agents SDK enables you to build agentic AI apps in a lightweight, easy-to-use package with very few abstractions. It's a production-ready upgrade of our previous experimentation for agents, Swarm. The Agents SDK has a very small set of primitives, agents, which are LLMs equipped with instructions and tools; handoffs, which allow agents to delegate to other agents for specific tasks; and guardrails, which enable the inputs to agents to be validated. In combination with Python, these primitives are powerful enough to express complex relationships between tools and agents, and allow you to build real-world applications without a steep learning curve. In addition, the SDK comes with built-in tracing that lets you visualize and debug your agentic flows, evaluate them, and even fine-tune models for your application.
Learn more
Swarm
Swarm is an experimental, educational framework developed by OpenAI to explore ergonomic, lightweight multi-agent orchestration. It is designed to be scalable and highly customizable, making it suitable for scenarios involving a large number of independent capabilities and instructions that are challenging to encode into a single prompt. Swarm operates entirely on the client side and, like the Chat Completions API it utilizes, does not store state between calls. This stateless nature allows for the construction of scalable, real-world solutions without a steep learning curve. Swarm agents are distinct from assistants in the assistants API; they are named similarly for convenience but are otherwise completely unrelated. It includes examples demonstrating fundamentals such as setup, function calling, handoffs, and context variables, as well as more complex scenarios like a multi-agent setup for handling different customer service requests in an airline context.
Learn more
Claude Agent SDK
The Claude Agent SDK is a developer toolkit that enables the creation of autonomous AI agents powered by Claude, allowing them to perform real-world tasks beyond simple text generation by interacting directly with files, systems, and tools. It provides the same underlying infrastructure used by Claude Code, including an agent loop, context management, and built-in tool execution, and is available for use in Python and TypeScript. With this SDK, developers can build agents that read and write files, execute shell commands, search the web, edit code, and automate complex workflows without needing to implement these capabilities from scratch. It maintains persistent context and state across interactions, enabling agents to operate continuously, reason through multi-step problems, take actions, verify results, and iterate until tasks are completed.
Learn more
Cua
Cua is a computer-use agent platform that lets AI agents see screens, click buttons, type, and run code just like a human across macOS, Windows, Linux, browsers, and mobile environments. It provides cloud-based, sandboxed desktops where agents can automate real software workflows without relying on APIs. Built on open-source Cua agents, the platform enables developers to build, run, and scale computer-use agents with precision and reliability. Cua supports multi-step tasks, structured outputs, and human-in-the-loop recovery for complex automation. Agents operate in fully isolated environments to ensure safety and reproducibility. Cua is designed to make AI interaction with real applications practical and scalable.
Learn more
