How to Use the AnythingLLM REST API

The AnythingLLM REST API gives you RAG in 3 minutes

AnythingLLM is a full-stack Retrieval-Augmented Generation (RAG) platform. It chunks your documents, embeds them into a vector database, and retrieves the most relevant context when you ask a question. Then it sends that context plus your query to an LLM for an answer.

The REST API lets you automate this entire pipeline:

• Upload documents to workspaces via /api/documents/upload
• Query documents with /api/workspaces/{id}/chat
• Manage users, LLM providers, and embeddings programmatically
• Receive webhook notifications when documents are indexed

Why AnythingLLM API beats ChatGPT API for document chat

ChatGPT API charges $0.01-0.05 per request and sends your data to OpenAI servers. AnythingLLM runs on your infrastructure. Your documents stay private. You pay $40/month flat (on Opsily) regardless of query volume—no per-request billing.

You also get vendor flexibility: switch between Ollama (local), OpenAI, Anthropic, or HuggingFace without reindexing documents. ChatGPT locks you in.

Production deployments need managed hosting

Self-hosting AnythingLLM means managing backups, security patches, database optimization, and SSL certificates. Opsily handles all of this. Deploy on Opsily in 3 minutes. Get automatic updates, GDPR-compliant German data centers, daily encrypted backups, and SSL—included.

Python example: Chat with your documents

import requests

BASE_URL = "https://your-instance.opsily.com/api"
API_KEY = "sk_..."
WORKSPACE_ID = "workspace_123"

headers = {"Authorization": f"Bearer {API_KEY}"}

# Chat with documents
response = requests.post(
    f"{BASE_URL}/workspaces/{WORKSPACE_ID}/chat",
    headers=headers,
    json={"message": "Summarize our privacy policy"}
)

print(response.json()["response"])  # LLM's answer based on your documents

JavaScript example: Embed in your web app

const chatWithDocuments = async (question) => {
  const response = await fetch(
    `https://your-instance.opsily.com/api/workspaces/workspace_123/chat`,
    {
      method: "POST",
      headers: {
        "Authorization": `Bearer ${API_KEY}`,
        "Content-Type": "application/json"
      },
      body: JSON.stringify({ message: question })
    }
  );
  const data = await response.json();
  return data.response;
};

Both examples use the same endpoints. Both work on Opsily managed hosting or self-hosted infrastructure. No vendor lock-in.