OpenAI has launched a new set of APIs and tools, making it easier for developers to build AI agents. The new Responses API combines the Chat Completions API with the Assistants API and can search the Internet and internal datasets.
The Responses API lets developers create AI agents powered by OpenAI’s LLMs. The new API should eventually replace the existing Assistants API, which helps developers build AI assistants within their apps. The Assistants API can perform queries using models, tools and files. On the other hand, the Responses API relies on the Chat Completions API, which is designed to generate a response based on a list of messages from a conversation.
Built-in tools
Another addition to the Responses API is web search functionality. This allows AI agents to retrieve current information from the Internet with source citations. This is available for the GPT-4o and GPT-4o-mini models. In turn, the file search tool makes it possible to extract relevant information from large amounts of documents, with support for various file types. That, in turn, is useful for searching a company’s internal data sets so that AI agents function based on the information a company wants. Together, the Web search and file search tools should ensure that the models and, thus, the AI agents function accurately.
OpenAI has announced that the current Assistants API will eventually be phased out, with a planned end date in mid-2026. The new Responses API should become the future direction for building agents on the OpenAI platform.
Tip: OpenAI comes out with ‘deep research’; the answer to DeepSeek?