In Assistants, Is there a limit to File size and number of files?

API
1

Assistant Questions :

  1. Is there a limit to file size and number of files that can be uploaded for an assistant? How is the upload charged(like embeddings?)
  2. Can we get reference of the source used? What reference(file) was used by assistant to answer that question? I see assistant , replying as 【24†source】, but what are they linked to , and how to access it?

Untitled
Untitled1079×212 167 KB

2

“You can attach a maximum of 20 files per Assistant, and they can be at most 512 MB each.”

OpenAI Platform.

1 Like
3

check the logs. they are linked to the file_ids from retrieval tool. which is which, not sure. it seems they chunk the files we uploaded. i also get 20† source but my uploaded files are less.

4

I’ve checked the logs, and matched it with the response. This text 【24†source】 is present in ‘plaintext’ form. And there are no ids to match it to. 20/24 , is not any ID anyway, as you said this is part of some chunking. And I want to know which chunk.

5

This is complete JSON, Please tell me where to map the source??

{
  "object": "list",
  "data": [
    {
      "id": "msg_kcxZu24jUOFM7oaKDUrZoE3F",
      "object": "thread.message",
      "created_at": 1699344897,
      "thread_id": "thread_56MMYcyjt65vscMXKQTbVRuM",
      "role": "assistant",
      "content": [
        {
          "type": "text",
          "text": {
            "value": "This is a dummy data【22†source】",
            "annotations": []
          }
        }
      ],
      "file_ids": [],
      "assistant_id": "asst_uotRuEcbBCiqISum6BTjlODW",
      "run_id": "run_mbZDjfOGBxU53oQrfgwOVHpM",
      "metadata": {}
    },
    {
      "id": "msg_rwhXkijXxmoD9AgHaXnbC7nd",
      "object": "thread.message",
      "created_at": 1699344864,
      "thread_id": "thread_56MMXdyjt64vscMYKQTbVRuO",
      "role": "user",
      "content": [
        {
          "type": "text",
          "text": {
            "value": "---This has been replaced---",
            "annotations": []
          }
        }
      ],
      "file_ids": [],
      "assistant_id": null,
      "run_id": null,
      "metadata": {}
    },
    {
      "id": "msg_ZU92qBkJqFXALDW7bLzGDHeM",
      "object": "thread.message",
      "created_at": 1699344764,
      "thread_id": "thread_56MMXcykt64vscMXKQTbVRuP",
      "role": "assistant",
      "content": [
        {
          "type": "text",
          "text": {
            "value": "---This has been replaced---",
            "annotations": []
          }
        }
      ],
      "file_ids": [],
      "assistant_id": "asst_totRuEcbBCiqISum6BTjlODZ",
      "run_id": "run_GucwBx4uh2AwBmI7lnK84oAb",
      "metadata": {}
    },
    {
      "id": "msg_S139O4rJ1h3hINyEGvT3T8L8",
      "object": "thread.message",
      "created_at": 1699344761,
      "thread_id": "thread_56MMXcyjt65vscMXKQVbVRuQ",
      "role": "user",
      "content": [
        {
          "type": "text",
          "text": {
            "value": "---This has been replaced---",
            "annotations": []
          }
        }
      ],
      "file_ids": [],
      "assistant_id": null,
      "run_id": null,
      "metadata": {}
    }
  ],
  "first_id": "msg_kcxZu24jUOFM7oaKDUrZoE3F",
  "last_id": "msg_S139O4rJ1h3hINyEGvT3T8L8",
  "has_more": false
}
6

The file_ids refer to the id of the files. I checked the Files list! Check the Files tab and check the ids of each of your files.

7

First let me first say I am somewhat new to this.
How can I take advantage of the 128K context if there is a 10,000 token limit per-minute at tier 1? Am I missing some something ? I do not plan to send a book up.

Second question is if the file size is limited to 512MB, can I break the file up and ask questions over multiple files ?

1 Like