Files - pre.dev

The files endpoints store documents for reuse across chat requests. Files are scoped to the workspace that uploaded them; another workspace’s file id returns 404.

Method	Path	Purpose
POST	`/v1/files`	Upload, `multipart/form-data`
GET	`/v1/files`	List the workspace’s files
GET	`/v1/files/{id}`	File metadata
GET	`/v1/files/{id}/content`	Download the bytes
DELETE	`/v1/files/{id}`	Delete

Upload

curl --fail-with-body https://api.pre.dev/v1/files \
  -H "Authorization: Bearer $PREDEV_API_KEY" \
  -F "purpose=user_data" \
  -F "file=@contract.pdf"

import fs from 'node:fs';
import OpenAI from 'openai';

const client = new OpenAI({ baseURL: 'https://api.pre.dev/v1', apiKey: process.env.PREDEV_API_KEY! });

const file = await client.files.create({
  file: fs.createReadStream('contract.pdf'),
  purpose: 'user_data',
});
console.log(file.id);

import os
from openai import OpenAI

client = OpenAI(base_url="https://api.pre.dev/v1", api_key=os.environ["PREDEV_API_KEY"])

with open("contract.pdf", "rb") as handle:
    uploaded = client.files.create(file=handle, purpose="user_data")
print(uploaded.id)

The upload takes the bytes in a file part and a purpose string (user_data for documents you will reference from chat messages) and returns the stored file object, including the id used in later requests.

List, fetch, download, delete

curl --fail-with-body https://api.pre.dev/v1/files \
  -H "Authorization: Bearer $PREDEV_API_KEY"

curl --fail-with-body "https://api.pre.dev/v1/files/$FILE_ID" \
  -H "Authorization: Bearer $PREDEV_API_KEY"

curl --fail-with-body "https://api.pre.dev/v1/files/$FILE_ID/content" \
  -H "Authorization: Bearer $PREDEV_API_KEY" -o contract.pdf

curl --fail-with-body -X DELETE "https://api.pre.dev/v1/files/$FILE_ID" \
  -H "Authorization: Bearer $PREDEV_API_KEY"

In the SDKs these are client.files.list(), client.files.retrieve(id), client.files.content(id), and client.files.delete(id) (Node client.files.del(id) on older versions).

Use a file in a request

Reference the uploaded file from a chat message as a file content part, and add the file-parser plugin so models without native document input can read it. See multimodal.