Skip to main content
POST

Introduction

The Responses API is OpenAI’s next-generation conversation interface, specifically designed for reasoning models (o-series, GPT-5 series) and advanced features. Compared to the traditional Chat Completions API, the Responses API provides more precise reasoning control, built-in tool support, and multimodal input capabilities.
gpt-6.1-sol: use Responses for function calls and for reasoning.effort="max". Chat Completions accepts requests without tools and supports reasoning_effort values low, medium, high, and xhigh. Neither interface accepts none or minimal. Omit temperature, top_p, and log-probability options. See GPT 6.1 Sol.

Use Cases

  • Reasoning-intensive tasks: Using reasoning models like o1, o3-mini, o4-mini, GPT-5
  • Web search requirements: Built-in Web Search Preview tool
  • Advanced tool calls: Support for Function Call and Custom Tool Call
  • Multi-turn conversation continuation: Conversation history management via previous_response_id

Authentication

Bearer Token, e.g., Bearer sk-xxxxxxxxxx

Request Parameters

string
required
Model identifier, e.g., gpt-5.5, o4-mini, o3-mini
string | array
required
Input text or an array of Responses input items.
integer
Maximum output tokens
boolean
Enable streaming output
object
Reasoning configuration, e.g., {"effort": "high", "summary": "detailed"}
array
Tool list, supports Web Search and function calls
string
Previous response ID for conversation continuation

Basic Examples

Advanced Features

Reasoning Control

Note: summary: "auto" automatically generates a reasoning summary, suitable for quick results.

Custom Function Calls

Responses function definitions place name, description, parameters, and strict beside type; do not wrap them in a function object. Handle output items by type. After running a function, return function_call_output with its call_id. For stateless requests, preserve the full output array, including any reasoning items, in the next input. On that follow-up, set tool_choice back to “auto” so the model can generate an answer.

Multimodal Input

Conversation Continuation

Use previous_response_id to maintain context across multi-turn conversations:

Response Format

Comparison: Responses API vs Chat Completions API