ChainThink reports that on September 11, OpenAI officially launched the public beta of its Agents API, making available to all developers the agent framework previously powering Codex and enterprise versions of ChatGPT.
Developers can build production-grade AI agents via the API, with pricing based on token consumption and tool usage.
This API supports long-running and complex tasks: developers can specify tasks, models, tools, and execution environments in a single call. Core capabilities include automatic context compression, parallel tool invocation, and multi-agent collaboration, enabling continuous operation for hours to days.
Customer cases show that after integrating SafetyKit, the cost per case decreased by 60%, the Hypha agent response failure rate dropped by 86%, Cirridae’s evaluation score increased from 0.71 to 0.85, and latency was reduced by fourfold.
On the infrastructure side, the Agents API uses the same sandbox environment as Codex and ChatGPT, and is integrated with partners including Blaxel, Cloudflare, DigitalOcean, Oracle, and Vercel, supporting various CPU, GPU, memory configurations, and deployment environments.
The official statement indicates that the Agents API is built on the open-source Codex execution framework and supports tools such as MCP, custom functions, and web search.
This launch marks Codex's expansion from an AI programming product into an agent infrastructure platform for developers.