OpenAI launches Agents API in public beta

OpenAI on Thursday released its Agents API in public beta, giving developers access to the same infrastructure that powers its Codex coding agent and ChatGPT for Work through a single API call. The launch was part of a broader wave of product releases from the company, which also introduced a Data agent for enterprise users and made its GPT-Live-1 voice model available to developers.

A Managed Harness for Long-Running Agents

The Agents API lets developers create cloud-based agents by specifying a task, model, tools, and compute environment, with OpenAI managing the underlying orchestration harness. Developers can run agents in OpenAI-hosted sandboxes, on their own infrastructure, or through partner integrations with Cloudflare ↗ 9.91%, DigitalOcean ↗ 2.52%, Oracle ↗ 0.82%, Vercel, and others.openai+1

Key technical features include automatic context compaction for sessions that stretch across hours, tool search to reduce token usage, programmatic parallel tool calling, and multi-agent support that delegates subtasks to subagents running concurrently. Jack Weissenberger, CTO of Ciridae, said in a statement that using the API raised the company's evaluation score to 0.85 from 0.71 and produced a fourfold latency reduction for subagent workflows.investing+1

The API carries no additional fees beyond standard token and tool usage pricing, and the harness is based on OpenAI's open-source Codex codebase.openai

Data Agent and Voice API Round Out the Day

Alongside the Agents API, OpenAI launched a Data agent inside ChatGPT Work that connects to enterprise data sources including Amazon Amazon.com, Inc. ↘ 0.35% Redshift, Google Alphabet Inc. ↗ 0.78% BigQuery, Snowflake ↗ 2.35%, Databricks, and MongoDB ↗ 2.71%, among others. The tool orchestrates context across data warehouses, Slack, file storage, and business intelligence dashboards to answer questions without requiring users to consolidate data first.venturebeat

"If you asked me a year ago what percentage of our data questions were getting answered end to end without needing anyone intervening, it was a very small fraction of the company," Arpan Shah, general manager of OpenAI's enterprise technology vertical, told VentureBeat. "Today, everyone at OpenAI can self-serve any data questions they have."venturebeat

OpenAI also made GPT-Live-1, its full-duplex voice model, available through its API. Yelp ↘ 5.17% announced it had integrated the model into its Yelp Host restaurant call-handling tool and its Hatch communications platform, which together have handled more than a million restaurant calls since launching in October 2025.stocktitan

Developer Access and What Comes Next

The Agents API is available to all developers starting Thursday, with OpenAI planning to iterate based on feedback before moving toward general availability. The release positions OpenAI to compete more directly with rival agent frameworks by offering a managed service that abstracts away harness maintenance, context management, and multi-agent orchestration — areas the company says it refined through operating Codex at scale.openai