Responses

Compact a conversation

POST /openai/v1/responses/compact

Compacts a conversation into a response object suitable for long-running and zero-data-retention scenarios.

Request Body

Content-Type: application/json

model string required

The model deployment to use for the compaction of this response.

input string | OpenAI.InputItem[] | null

Any of the following:

string

OpenAI.InputItem[]

previous_response_id string | null

instructions string | null

prompt_cache_key string | null

prompt_cache_retention enum

in_memory24h

prompt_cache_options OpenAI.PromptCacheOptionsParam | null

Options for prompt caching. Supported for gpt-5.6 and later models. By default, OpenAI automatically chooses one implicit cache breakpoint. You can add explicit breakpoints to content blocks with prompt_cache_breakpoint. Each request can write up to four breakpoints. For cache matching, OpenAI considers up to the latest 80 breakpoints in the conversation, without a content-block lookback limit. Set mode to explicit to disable the implicit breakpoint. The ttl defaults to 30m, which is currently the only supported value. See the prompt caching guide for current details.

ttl enum

30m

mode enum

implicitexplicit

service_tier enum

autodefaultfastflexpriority

Responses

The request has succeeded.

Body

Content-Type: application/json

id string required

The unique identifier for the compacted response.

object enum required

The object type. Always response.compaction.

response.compaction

output OpenAI.ItemField[] required

The compacted list of output items.

created_at integer required

Unix timestamp (in seconds) when the compacted conversation was created.

usage OpenAI.ResponseUsage required

Represents token usage details including input tokens, output tokens, a breakdown of output tokens, and the total tokens used.

input_tokens integer required

input_tokens_details OpenAI.ResponseUsageInputTokensDetails required

cached_tokens integer required

cache_write_tokens integer required

output_tokens integer required

output_tokens_details OpenAI.ResponseUsageOutputTokensDetails required

reasoning_tokens integer required

total_tokens integer required

Request

curl -X POST https://api.openai.com/v1/responses/compact \
    -H "Content-Type: application/json" \
    -H "Authorization: Bearer $OPENAI_API_KEY" \
    -d '{
      "model": "gpt-5.1-codex-max",
      "input": [
        {
          "role": "user",
          "content": "Create a simple landing page for a dog petting café."
        },
        {
          "id": "msg_001",
          "type": "message",
          "status": "completed",
          "content": [
            {
              "type": "output_text",
              "annotations": [],
              "logprobs": [],
              "text": "Below is a single file, ready-to-use landing page for a dog petting café:..."
            }
          ],
          "role": "assistant"
        }
      ]
    }'

import OpenAI from "openai";

const openai = new OpenAI();

// Compact the previous response if you are running out of tokens
const compactedResponse = await openai.responses.compact({
  model: "gpt-5.1-codex-max",
  input: [
    {
      role: "user",
      content: "Create a simple landing page for a dog petting café.",
    },
    // All items returned from previous requests are included here, like reasoning, message, function call, etc.
    {
      id: "msg_030d085c0b53e67e0069332e3a72d4819c96c6f2c4adc15d33",
      type: "message",
      status: "completed",
      content: [
        {
          type: "output_text",
          annotations: [],
          logprobs: [],
          text: "Below is a single file, ready-to-use landing page for a dog petting café:...",
        },
      ],
      role: "assistant",
    },
  ],
});

// Pass the compactedResponse.output as input to the next request
console.log(compactedResponse);

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ.get("OPENAI_API_KEY"),  # This is the default and can be omitted
)
compacted_response = client.responses.compact(
    model="gpt-5.4",
)
print(compacted_response.id)

import OpenAI from 'openai';

const client = new OpenAI({
  apiKey: process.env['OPENAI_API_KEY'], // This is the default and can be omitted
});

const compactedResponse = await client.responses.compact({ model: 'gpt-5.4' });

console.log(compactedResponse.id);

package main

import (
    "context"
    "fmt"

    "github.com/openai/openai-go"
    "github.com/openai/openai-go/option"
    "github.com/openai/openai-go/responses"
)

func main() {
    client := openai.NewClient(
        option.WithAPIKey("My API Key"),
    )
    compactedResponse, err := client.Responses.Compact(context.TODO(), responses.ResponseCompactParams{
        Model: responses.ResponseCompactParamsModelGPT5_4,
    })
    if err != nil {
        panic(err.Error())
    }
    fmt.Printf("%+v
", compactedResponse.ID)
}

package com.openai.example;

import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;
import com.openai.models.responses.CompactedResponse;
import com.openai.models.responses.ResponseCompactParams;

public final class Main {
    private Main() {}

    public static void main(String[] args) {
        OpenAIClient client = OpenAIOkHttpClient.fromEnv();

        ResponseCompactParams params = ResponseCompactParams.builder()
            .model(ResponseCompactParams.Model.GPT_5_4)
            .build();
        CompactedResponse compactedResponse = client.responses().compact(params);
    }
}

require "openai"

openai = OpenAI::Client.new(api_key: "My API Key")

compacted_response = openai.responses.compact(model: :"gpt-5.4")

puts(compacted_response)

Response

{
  "id": "resp_001",
  "object": "response.compaction",
  "created_at": 1764967971,
  "output": [
    {
      "id": "msg_000",
      "type": "message",
      "status": "completed",
      "content": [
        {
          "type": "input_text",
          "text": "Create a simple landing page for a dog petting cafe."
        }
      ],
      "role": "user"
    },
    {
      "id": "cmp_001",
      "type": "compaction",
      "encrypted_content": "gAAAAABpM0Yj-...="
    }
  ],
  "usage": {
    "input_tokens": 139,
    "input_tokens_details": {
      "cached_tokens": 0
    },
    "output_tokens": 438,
    "output_tokens_details": {
      "reasoning_tokens": 64
    },
    "total_tokens": 577
  }
}