OpenAI announced new features for developers, including the Responses API, which unifies various functionalities for easier integration of text and image generation, web search, and more. Additionally, three new tools were introduced: a web search tool for real-time data, a file search tool for querying uploaded files, and a computer use tool for automating tasks, along with observability tools for monitoring API calls.
In a recent video stream, OpenAI announced several new features aimed at developers, filling a gap in their API offerings that previously focused more on consumers and enterprises. The announcement included the introduction of the Responses API, built-in tools, and an observability tool for monitoring API calls. The video highlighted the evolution of OpenAI’s APIs, noting that the Responses API aims to unify various functionalities, making it easier for developers to integrate features like text generation, image generation, and web search into their applications.
The Responses API represents a significant advancement, combining elements from previous APIs, such as the completions and chat APIs. This new API allows for a more streamlined approach to input and output, supporting various types of content, including text, images, and function calls. The goal is to simplify the developer experience while providing access to a broader range of capabilities, including reasoning models and streaming responses. The video emphasized that while the chat completions API will remain available, the assistant API is set to be deprecated by mid-2026.
Three new tools were introduced alongside the Responses API. The first is the web search tool, which allows developers to integrate real-time search capabilities into their applications. This tool is designed to enhance the intelligence of language models by sourcing up-to-date information from the internet. The video demonstrated how the web search tool works, including its pricing structure, which starts at $30 per 1,000 calls, making it accessible for developers looking to leverage real-time data.
The second tool is file search, which enables users to upload files and perform queries over them. This revamped feature is positioned as a general retrieval-augmented generation (RAG) solution, allowing for easy integration of file data into applications. The video noted that this tool could potentially compete with existing vector store solutions by offering a straightforward way to manage and query uploaded files, complete with metadata and citation capabilities.
Lastly, the video discussed the computer use tool, which powers OpenAI’s agent capabilities. This tool allows users to automate tasks by interacting with web browsers, although it is currently limited to the ChatGPT Pro Plan and available only in the United States. The video concluded by mentioning the new observability tools that OpenAI has introduced, which enable developers to trace and monitor their API calls and agent workflows. Overall, the announcements signal OpenAI’s commitment to empowering developers in the growing field of AI agents, with further exploration of the agents SDK planned for a future video.