Chat

Chat with document collections and AI agents.

Retrieve user chats

GEThttps://www.openindex.ai/api/chat

Query parameters

Response

200

A list of chats.

Body

application/json

okboolean

chatsarray of Chat (object)

Request

JavaScriptCurl

const response = await fetch('https://www.openindex.ai/api/chat', {
    method: 'GET',
    headers: {},
});
const data = await response.json();

Response

{
  "ok": false,
  "chats": [
    {
      "id": "text",
      "userId": "text",
      "agentId": "text",
      "collectionId": "text",
      "status": "text",
      "current": [],
      "tools": [
        {
          "id": "text",
          "checked": false
        }
      ]
    }
  ]
}

Send a message to an AI agent or collection.

POSThttps://www.openindex.ai/api/chat

Body

application/json

chatIdstring

Enter null to create a new conversation or an existing chatId to continue a conversation

agentIdstring

collectionIdstring

messagestring

streamingboolean

toolsarray of Tool (object)

modelModelName (enum)

Name of the LLM model to use.

gpt-3.5-turbogpt-3.5-turbo-16kgpt-4palm-2llama-2-13b-chatllama-2-70b-chatcodellama-13b-instructcodellama-13b-pythonmistral-7bllava-13bfuyu-8b

lastUploadedFilestring

Publicly accessible URL of the last uploaded file. Typically used to feed an image to a multi-modal model.

languagestring

2-letter ISO code of the language the model should use to reply.

Response

200403500

The response to the message from the language model.

Body

application/json

okboolean

messagestring

errorstring

sourcesarray of string

followupobject

contextobject

chatIdstring

agentIdstring

collectionIdstring

Request

JavaScriptCurl

const response = await fetch('https://www.openindex.ai/api/chat', {
    method: 'POST',
    headers: {
      "Content-Type": "application/json"
    },
    body: JSON.stringify({}),
});
const data = await response.json();

Response

{
  "ok": false,
  "message": "text",
  "error": "text",
  "sources": [
    "text"
  ],
  "chatId": "text",
  "agentId": "text",
  "collectionId": "text"
}

Please see the list of supported models for the model parameter above.

If you pass true to the parameter streaming, the API will open a stream returning a JSON Lines format (i.e. one json object per line).