OpenAI has introduced a new developer tool called the Responses API, designed to simplify and enhance the creation of advanced AI agents. The tool, which replaces the Assistants API, aims to address key challenges developers face when building AI systems, such as managing multi-turn interactions, integrating hosted tools, and handling multiple data types like text, images, and audio.  

According to an official blog post by OpenAI, the Responses API is "stateful", meaning it automatically manages conversation history, eliminating the need for developers to manually track interactions. This allows developers to retrieve full conversation contexts at any time, making it easier to build complex, multi-turn AI applications. Additionally, the API supports hosted tools like web search and file search, enabling AI agents to perform tasks such as retrieving web results or analysing documents without requiring manual intervention.  

For example, developers can now build AI systems that analyse an image, perform a web search, and summarise the results in a single API call. This contrasts with older APIs, which often require multiple steps and separate requests to achieve the same outcome, states the OpenAI blog post. The Responses API also supports multimodal inputs, allowing developers to create AI agents that can process and respond to text, images, and audio seamlessly.  

The tool is available to all developers at no extra cost and is part of OpenAI's efforts to stay competitive in the rapidly evolving AI landscape. The release comes amid growing competition from Chinese AI startups like DeepSeek and Monica, which have recently launched cost-efficient models and autonomous AI agents.  

OpenAI plans to phase out the Assistants API by the second half of 2026, encouraging developers to transition to the Responses API.