OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
RAG & Knowledge Systems

Vision is All You Need: V-RAG Demo

Vision is All You Need is an open-source visual retrieval-augmented generation demo for searching PDF documents and answering questions using retrieved page images. It converts pages to images, embeds them with ColPali, stores vectors in Qdrant, and sends matching pages with the query to GPT-4o or GPT-4o-mini. The project includes a FastAPI API and React frontend, with Modal documented for serving and deployment.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Vision is All You Need: V-RAG Demo — OpenViST project profile in RAG & Knowledge Systems.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it to explore visual retrieval and question-answering over PDFs when preserving page-level visual information is useful.

Who it is for

  • Developers exploring visual RAG for PDF collections.
  • Teams prototyping question-answering over image-heavy reports, forms or manuals.

Documented uses

  • Index PDF pages as image vectors.
  • Retrieve pages relevant to a natural-language query.
  • Generate an answer based on the query and retrieved page images.
  • Prototype an API and frontend workflow for visual document search.
Before you choose

Requirements & limits

The documented setup requires a Hugging Face account and login, plus an OpenAI API key. Local frontend development requires Node.js.

  • The demo embeds page images directly rather than using a text-chunking workflow.
  • PDF is the documented input; other image-renderable documents are described as possible in theory but would require conversion and testing.
  • Indexing uses an in-memory vector database in the documented API workflow, and processing time varies with PDF size and GPU.
  • The repository lists an MIT license.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Run the V-RAG demo with Modal

Serve the demo locally through Modal after configuring its documented credentials.

Before you start

  • Python 3.11 or higher
  • A Hugging Face account and login using transformers-cli
  • An OpenAI API key
  • Modal installed

Steps

  1. Add OPENAI_API_KEY and HF_TOKEN to the dotenv file.

  2. Install Modal with pip install modal.

  3. Run modal setup.

  4. Run modal serve \main.py.

Expected outcome

Modal serves the demo and provides a URL for accessing its API.

Guide limitations

  • The README notes that PDF indexing time depends on document size and the GPU used; its current demo uses an A10G GPU.
Documentation 1 ↗
Source-based practical guide

Index a PDF and search it through the API

Add a PDF to the demo’s in-memory vector database and query its indexed pages.

Before you start

  • A running demo and the URL provided by Modal
  • A PDF file

Steps

  1. Open the Modal URL with /docs appended.

  2. Use POST /collections, choose Try it out, upload a PDF and execute the request.

  3. After indexing, use POST /search to submit a query.

Expected outcome

The API searches similar page images and returns a response generated using the query and matching images.

Guide limitations

  • The documented vector database for this API workflow is in memory.
  • Indexing takes time depending on PDF size and the GPU used.
Documentation 1 ↗
COMPARE OPTIONS

Compare options for Retrieval-augmented applications

These projects document a common task and share a project role. This is a source-based comparison, not an execution benchmark or a guarantee of interchangeability.

ProjectDocumented purposePlatformAccess and requirements
Vision is All You Need: V-RAG DemoIndex PDF pages as image vectors.Cloud, Self-hosted, WebThe documented setup requires a Hugging Face account and login, plus an OpenAI API key. Local frontend development requires Node.js.
Rag Powered AI Assistant Transforming Data Retrieval And Analysis Across The Web And PdfsRetrieve relevant passages from web and PDF content to answer a question.Web
LangChain in Supabase Edge FunctionsInitialize and start a Supabase project locally.Cloud, Server
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
14 October 2026
Original website title
Softlandia-Ltd/vision-is-all-you-need · Official homepage title · retained in its source language
How we review sourcesSuggest a correction