What is Responses API
Responses API is the core interface launched by OpenAI for building AI Agents. Responses API is an upgraded version of the Chat Completions API, combining the tool-calling capabilities of the Assistants API to support more complex task processing. Based on a single call, developers use multiple built-in tools (such as web search, file search, computer use, etc.) and multi-turn model interactions to complete complex tasks. Responses API simplifies the project development process, lowers the technical threshold, and supports streaming events to improve response efficiency.
Main features of Responses API
- Integrated multiple tool-calling capabilities: Combining the conversation generation capabilities of the Chat Completions API and the tool-calling functionality of the Assistants API, it supports developers to combine multiple tools (such as web search, file search, computer use, etc.) in a single call to complete complex tasks.
- Supports streaming event processing: Provides predictable streaming event support, processes and responds to data streams in real time, and improves interaction efficiency and user experience.
- Monitoring and debugging support: Integrates with OpenAI's monitoring tools (such as Logs and Traces) to help developers track and debug Agent workflows and optimize performance and stability.
- Unified Item-based design: User input, model output, and tool calls are all abstracted into Items with a unified structure. The structure is clear and highly extensible, laying a good foundation for feature expansion and integration.
- Simpler polymorphism handling: Based on the Item-based design and intelligent type inference, it simplifies the complexity of polymorphism handling, allowing different types of Items to be handled without a large amount of conditional judgment code.
Built-in tools of Responses API
- Web Search: Quickly obtain the latest answers, supports gpt-4o and gpt-4o-mini models, and provides relevant citations.
- File Search: Retrieves relevant information from a large number of documents, supports multiple file types and query optimization.
- Computer Use: Uses the Computer-Using Agent (CUA) model to support automated tasks, such as browser workflows.
Official website address of Responses API
- Official website address：https://platform.openai.com/docs/quickstart?api-mode=responses
Product pricing of Responses API
- Token fees: Responses API charges based on the number of Tokens generated and processed.
- Tool usage fees：
- Web Search: The pricing for using the gpt-4o-search-preview and gpt-4o-mini-search-preview models is 30 USD and 25 USD per thousand queries, respectively.
- File Search: 2.50 USD per thousand queries, file storage is priced at 0.10 USD/GB/day, and the first GB is free.
- Computer Use: 3 USD per 1 million input Tokens and 12 USD per 1 million output Tokens.
Application scenarios of Responses API
- Intelligent customer service: Automatically answer customer questions, process orders and refunds.
- Market research: Collect data from the web and documents for analysis.
- Content creation: Write articles, edit documents, and publish social media updates.
- Sales support: Find potential customers and generate sales proposals.
- Personalized Education: Provide customized learning materials and evaluate student performance.