How do I get every message in a thread?

I’m using the JS sdk and my code looks like this:


This only returns 20 messages. I see that I can pass in a “before” option but it’s unclear what id I need to pass in. If I pass in message ids, nothing gets returned

I figured it out. MessagesPage (the return of the .list call) has a function called getNextPage on it. I just have to throw that call in a while loop that checks if the data length is greater than 0

I don’t know what a “JD SDK” is, but here’s how you use the openai library if you don’t want to just add query strings to the URL…

Using Pagination to Accumulate Messages in a Thread with the OpenAI Python Library

The OpenAI Python library provides a convenient way to list messages in a thread using the list method. When dealing with threads that contain hundreds of messages, pagination is essential. This guide will show you how to use the pagination parameters to retrieve all messages in a thread, step by step.


First, ensure that you have the latest OpenAI Python library installed:

pip install --upgrade openai

Example Usage

Here’s an example of how to use the list method without any parameters:

from openai import OpenAI

client = OpenAI()
thread_messages = client.beta.threads.messages.list("thread_abc123")

To accumulate hundreds of messages, you’ll need to use the pagination parameters. The key parameters for pagination are after, before, and limit.

Pagination Parameters

  • after: A cursor for use in pagination. It specifies the object ID that defines your place in the list. For instance, if you make a list request and receive 100 objects ending with obj_foo, your subsequent call can include after=obj_foo to fetch the next page.
  • before: Similar to after, but it specifies the object ID to fetch the previous page.
  • limit: The number of objects to return per page. The limit can range between 1 and 100, with the default being 20.

Retrieving All Messages

Here is a step-by-step example of how to retrieve all messages in a thread using the after parameter for pagination:

from openai import OpenAI

client = OpenAI()
thread_id = "thread_abc123"
all_messages = []
limit = 100  # Maximum allowed limit per request
after = None

while True:
    response = client.beta.threads.messages.list(thread_id, limit=limit, after=after)
    messages =
    if not messages:
    after = messages[-1].id  # Set the 'after' cursor to the ID of the last message

print(f"Total messages retrieved: {len(all_messages)}")

Detailed Breakdown

  1. Initialization: Initialize the OpenAI client and set the thread_id for the thread you want to fetch messages from.
  2. Pagination Loop: Use a while loop to continuously fetch pages of messages until no more messages are returned.
  3. Fetch Messages: Call the list method with the limit and after parameters. Initially, after is None.
  4. Extend List: Add the fetched messages to the all_messages list.
  5. Update Cursor: Update the after cursor to the ID of the last message in the current batch.
  6. Break Condition: Exit the loop when no more messages are returned.

Handling Edge Cases

  • Rate Limiting: Be mindful of API rate limits. You may need to add delays between requests or handle rate limit errors gracefully.
  • Error Handling: Implement error handling to manage network issues or invalid responses.

This approach ensures that you can accumulate hundreds (or even thousands) of messages from a thread efficiently. Adjust the limit parameter as needed based on your specific requirements.

The above is AI after I provided it all the documentation and source. It doesn’t understand parsing out of pydantic well, though.