OpenAI Launches New Tools & SDK to Simplify AI Agent Development

OpenAI has announced a new suite of APIs, built-in tools, and an open-source SDK designed to simplify the development of useful, reliable AI agents for developers and enterprises. These new releases provide the essential building blocks to move AI agents from concept to production, helping automate complex, multi-step tasks across industries.

Streamlined APIs and Tools for Agentic Applications

The highlight of the release is the Responses API, which merges the simplicity of OpenAI’s Chat Completions API with the tool-use capabilities of the Assistants API. The Responses API allows developers to integrate multiple tools—such as web search, file search, and computer use—into a single, unified API call, eliminating the need for extensive prompt engineering or complex orchestration logic.

Key features include:

• Web Search Tool: Provides real-time answers with clear citations, ideal for research assistants, shopping agents, or market intelligence tools. Web search in the API is powered by the same model used for ChatGPT search, achieving 90% and 88% accuracy on the SimpleQA benchmark for factual question answering. The tool is available to all developers in preview via the Responses API, with direct access to fine-tuned search models also offered through the Chat Completions API (gpt-4o-search-preview and gpt-4o-mini-search-preview). Pricing starts at $30 and $25 per thousand queries, respectively.

• File Search Tool: Allows developers to efficiently retrieve relevant information from large document sets across multiple file types, with built-in support for query optimization, metadata filtering, and custom reranking. Use cases include enhancing customer support agents, legal assistants, and technical documentation queries. For example, Navan uses file search to power its AI travel agent, enabling users to get precise answers from internal knowledge bases like travel policies. The tool supports a streamlined RAG (retrieval-augmented generation) pipeline without extra configuration.

  This tool is available to all developers in the Responses API. Pricing starts at $2.50 per thousand queries and $0.10/GB/day for file storage, with the first GB free. It remains available in the Assistants API, and developers can also use the new search endpoint in Vector Store API objects to query their data directly for broader application use.
  
  

• Computer Use Tool: Enables developers to automate browser-based and desktop workflows by translating model-generated mouse and keyboard actions into executable commands. Powered by OpenAI’s Computer-Using Agent (CUA) model—the same model used to power Operator—it supports tasks like quality assurance testing, data entry, and research workflows on systems without API access. The model has achieved state-of-the-art benchmarks, with an 87% success rate on WebVoyager and 58.1% on WebArena, though it remains less reliable for full computer use tasks (38.1% on OSWorld), making human oversight essential.

• This tool is available as a research preview in the Responses API for developers in usage tiers 3-5. Pricing starts at $3 per 1M input tokens and $12 per 1M output tokens. OpenAI has also integrated robust safety measures, including confirmation prompts, prompt injection checks, and environment isolation tools, to mitigate risk.

The Responses API also includes improved observability features, allowing developers to trace, evaluate, and optimize agent workflows while ensuring data privacy—business data is not used to train OpenAI’s models by default.

What About Existing APIs?

For developers already using OpenAI’s Chat Completions API or Assistants API, the new Responses API builds on familiar capabilities. It combines the simplicity of Chat Completions with the tool-use flexibility of Assistants, making it a more powerful, unified option for developing agentic applications. While Chat Completions remains widely supported, and the Assistants API will continue to receive updates, OpenAI plans to phase out the Assistants API by mid-2026, providing a clear migration path to the Responses API.

Agents SDK Simplifies Multi-Agent Orchestration

Alongside the APIs, OpenAI introduced the Agents SDK, an open-source framework designed to orchestrate single-agent and multi-agent workflows with built-in guardrails, handoffs, and observability tools. It works seamlessly with the Responses API and other chat-style APIs, offering developers flexible, production-ready agent solutions.

The Agents SDK is suitable for a wide range of real-world use cases, from automating customer support and streamlining multi-step research, to powering content creation, code reviews, and sales prospecting. Real-world examples include:

• Coinbase: Leveraged the Agents SDK to prototype AgentKit, integrating AI agents with crypto wallets and on-chain activities.

• Box: Quickly built agents that allow enterprises to securely search and extract insights from both proprietary data and public web sources.

Python support is available immediately, with Node.js compatibility coming soon.

Looking Ahead: Building the Future of AI Agents

OpenAI sees agents as integral to the future workforce, empowering industries with intelligent, autonomous systems. These new tools reflect OpenAI’s broader strategy to make agent development more accessible, reliable, and scalable, accelerating adoption across sectors.

With further updates and deeper integrations expected in the coming months, OpenAI is laying the groundwork for a seamless platform experience—giving developers everything they need to deploy high-performing AI agents that deliver real-world impact.

Editor’s Note: This article was created by Alicia Shapiro, CMO of AiNews.com, with writing, image, and idea-generation support from ChatGPT, an AI assistant. However, the final perspective and editorial choices are solely Alicia Shapiro’s. Special thanks to ChatGPT for assistance with research and editorial support in crafting this article.