OpenAI Agents API is a managed version of the Codex harness that lets developers build durable cloud agents with a single API call instead of building their own orchestration. It handles long sessions, smart tool use, and subagents working in parallel, and runs on OpenAI or partner sandboxes. The harness is open source and available in public beta, with no extra fees beyond token and tool usage.
OpenAI Agents API
This version has not been reviewed by an editor.
Report incorrect information
Signed-in users can report a broken link, incorrect information or duplicate entry. An editor will review it.
Cloud agents, run on OpenAI's Codex harness
OpenAI Agents API Introduction
Key Features
- Managed Codex harness for building cloud agents with one API call
- Support for long-running agent sessions
- Smart tool use orchestration
- Subagents working in parallel
- Runs on OpenAI or partner sandboxes
- Open source harness
Use Cases
- Building durable cloud agents without custom orchestration
- Running parallel subagents for complex tasks
- Automating multi-step developer workflows
Why Startups Use It
Startups can ship agent-based products quickly without investing in orchestration infrastructure, since the API manages sessions, tool use, and parallel subagents while charging only for token and tool usage.
Alternative Options
LangChain, CrewAI, AutoGen
Frequently Asked Questions
What is the OpenAI Agents API?
It is OpenAI's managed version of the Codex harness, enabling developers to build durable cloud agents with a single API call.
Does it cost extra beyond API usage?
No, there are no extra fees beyond token and tool usage.
Is the harness open source?
Yes, the harness is open source and the product is currently in public beta.
Where do agents run?
Agents run on OpenAI or partner sandboxes.
More About OpenAI Agents API
PricingPaid
Listed DateSep 16, 2026
Authority Badge
Add our badge to your website to showcase product credibility and listing status.
Featured List