On September 10 (local time), OpenAI announced the launch of the public beta for its Agents API, allowing developers to build and run cloud-based agents through API calls. The API reuses the agent execution framework and infrastructure of Codex, enabling developers to specify tasks, models, tools, and runtime environments, and choose to deploy them in OpenAI-hosted sandboxes, their own infrastructure, or partner sandboxes. Currently, the API is available for public beta testing to all developers without additional charges, with users only required to pay based on the actual usage of model tokens and tool calls.
