# Hugging Face AI Sheets

> Open-source no-code spreadsheet tool for building, enriching, and transforming datasets using AI models.

Hugging Face AI Sheets is an open-source tool from Hugging Face for building, enriching, and transforming datasets with AI models and no code. It can be tried in a Hugging Face Space, run locally with Docker or pnpm, or deployed on the Hub. It is released under the Apache 2.0 license.

## What It Is

AI Sheets is a spreadsheet-style application for dataset work. Columns are populated by AI models, with thousands of open models from the Hugging Face Hub available via Inference Providers, or local and custom models. The README mentions gpt-oss from OpenAI as one supported model.

## Running and Deployment

Users can open the hosted Space, start the app with a Docker image using a Hugging Face token, or clone the repository and run it with pnpm. A production build can be served through the built-in Express server. For larger datasets, the repository documents scripts run through HF Jobs, using either the inference client or vLLM, driven by a YAML config that holds the prompts.

## Custom and Local Models

By default, inference uses Hugging Face Inference Providers. Any model endpoint that follows the OpenAI API specification, such as Ollama running locally, can be used by setting MODEL_ENDPOINT_URL and MODEL_ENDPOINT_NAME. Models from Inference Providers remain selectable in the column settings. Text-to-image generation cannot yet be customized and always uses Inference Providers.

## Configuration

Environment variables control authentication (OAuth or HF_TOKEN), default model and provider, organization billing, concurrent request limits, the data directory, optional Serper web search, and telemetry. The app is built with Qwik and QwikCity in TypeScript.

## Features
- No-code dataset building, enrichment, and transformation with AI models
- Access to thousands of open models via Hugging Face Inference Providers
- Support for local and custom LLMs through OpenAI-compatible endpoints
- Run locally with Docker or pnpm, or use the hosted Space
- Large-scale dataset generation scripts via HF Jobs with inference client or vLLM
- Hugging Face OAuth or token authentication
- Optional web search via Serper
- Configurable default model, provider, and concurrency
- Text-to-image generation via Inference Providers

## Integrations
Hugging Face Hub, Hugging Face Inference Providers, HF Jobs, Ollama, vLLM, Serper, OpenAI-compatible APIs, Docker

## Platforms
WEB, API, VSC_EXTENSION, CLI

## Pricing
Open Source

## Links
- Website: https://huggingface.co/spaces/aisheets/sheets
- Documentation: https://github.com/huggingface/aisheets#readme
- Repository: https://github.com/huggingface/aisheets
- EveryDev.ai: https://www.everydev.ai/tools/hugging-face-ai-sheets