Documentation | AI Box
NEWConnect to other AI tools with an MCP
Documentation

Everything you need to build with AI Box

Read by surface (Playground, Builder, Marketplace) or by concept (MCP, models). The most useful place to start is the first section on the left.

Get started

Getting started

AI Box is one subscription that gives you 80+ AI models plus a no-code workspace for chaining them into reusable workflows. This page walks you through what to try first.

1 · Open the Playground

Pick any model from the dropdown and start chatting. Same chat box works across GPT, Claude, Gemini, Grok, Llama, Mistral, Sonar, and 70+ others. No API keys.

2 · Compare models side-by-side

After you prompt a model, hit Compare to fan that same prompt out across up to six others. Every response opens in its own column so you can scan them side by side.

3 · Build your first workflow

Open the Builder, describe the tool you want, and AI Box links the right models together for you. Or drag nodes onto the canvas yourself if you prefer manual control.

4 · Publish to the Marketplace

Workflows you build can be published so others can run them. You earn credits whenever someone else uses one of your tools.

Surfaces

Playground

The Playground is a multi-model chat surface. Use it for one-off questions, prompt iteration, and side-by-side comparisons before committing a model into a workflow.

Switch models mid-thread

Change the model dropdown at any point. Your conversation history travels with you, so the new model sees what was said before.

Attach files

Drop in PDFs, images, spreadsheets, or audio. The model reads them in context. No copy-paste needed.

Compare mode

Once you have a prompt going, hit Compare to re-run it across up to six other models. Each response opens in its own column so you can scan them side by side.

Surfaces

Builder

The Builder is a drag-and-drop canvas for chaining models, file inputs, and outputs into reusable workflows. Describe what you want or build nodes by hand.

Describe to build

Type a sentence ('a weekly newsletter generator for wellness brands') and AI Box scaffolds the right nodes + connections for you.

Add nodes manually

Double-click empty canvas to add an AI model node. Drag from output to input handles to connect them.

Design + Format

Once the workflow is built, the Design tab lets you customize what the end-user sees: input fields, branding, output layout.

Publish

Workflows ship as standalone tools other users can run via a link, embed, or your Marketplace listing.

Surfaces

Marketplace

The Marketplace is a public catalog of workflows people have built. Find ready-made tools, run them, fork them, or publish your own.

Browse + run

Every listing has a live preview. Hit Run to use it with your own inputs without leaving the page.

Fork and edit

Click Fork on any listing to drop its workflow into your Builder, where you can customize prompts and models.

Publish your own

Once a workflow is working, hit Publish from the Builder to add it to the Marketplace under your account.

Concepts

MCP server

Every AI Box workflow is exposed as a Model Context Protocol (MCP) server. That means Claude, ChatGPT, Gemini, or any MCP-capable client can call your workflows from their own chat.

What MCP is

MCP is an open standard that lets an LLM client discover and call external tools at runtime. AI Box implements it server-side so your workflows show up as callable tools inside other AI clients.

Connect a client

Grab your MCP URL from Settings → MCP, paste it into Claude/ChatGPT/Gemini's MCP server config, and your workflows appear in that client's tool picker.

Authentication

Each MCP connection is scoped to a personal access token. Generate one from Settings → API, rotate at any time.

Concepts

Models

AI Box supports 80+ models across text, image, audio, and video. They're available in one subscription. No provider keys, no per-model billing.

Text

GPT-5, GPT-4o, Claude 4 Opus, Claude 4 Sonnet, Gemini 2.5 Pro, Grok 3, Llama 4, Mistral Large, DeepSeek R1, Sonar Pro, and 20+ more.

Image

Flux 1.1 Pro, GPT Image, Stable Diffusion XL, Ideogram, HiDream, RunDiffusion, DreamShaper, and others.

Audio

ElevenLabs (TTS + music), Whisper (transcription), OpenAI TTS, and more.

Video

ByteDance Seedance, PixVerse, and a growing list.

Still stuck?

We answer every support email within one business day.

support@aibox.ai