Start Here

Free, Pro, or your own instance

Free and Pro run on big-agi.com, signed in; Open is the build you deploy yourself. What each includes, who bills you, and whether to keep the subscription you already pay.

Browse documentation

Free and Pro run on big-agi.com, signed in: Free is the product, Pro adds a subscription on top. Open is the same core, MIT-licensed, deployed and updated by you. The list of AI services is identical in all three.

What each tier includes

Open - you host itFreePro
800+ models, on your keys New models land here first
Beam - several models on one question
AI Inspector - the literal request, as sent
Personas-
Rambles-
Browsing, web search, premium voiceYou supply and run them
Account, cloud backup, sync-Account only - no backup, no sync1 GB: personas and chats, unlimited devices
Email support from the builder--
Pro

Chats, personas and rambles stay on this device unless you subscribe to Pro: cloud backup and multi-device sync (1 GB). On Open there is no sync. Disabling sync never deletes local data.

Teams arrange the Business plan by contact on big-agi.com.

Who bills you

Two bills, and only one is ours.

BillWho charges youFor what
Model usageYour AI service, directlyEvery request, priced per token
Big-AGI ProBig-AGI, $10.99/moBackup, sync, 1 GB, email support

Big-AGI never meters, caps, or takes a cut of model usage. Host Open and you pay whoever runs the machine, nothing to us.

The key behind the first bill comes from an AI service account you create, not the service's consumer chat account. Fund it there, by card or prepaid credit. Cap the spend in the service's console: connect an AI service, keep your key safe.

Keeping the subscription you already pay

If your subscription works for you, keep it. ChatGPT, Claude and Gemini give a lot away free, and their paid plans are subsidised - good deals for what they cover.

Big-AGI runs on API keys, not on those subscriptions: connect an AI service walks the console. Consider it when you want several models on one question, vendors side by side, or a local model next to the cloud ones. The consumer extras - voice on a phone, memory, connectors, research agents - have no equivalent here. Feature by feature: the comparison page.

Switching later

Chats and settings move in every direction as one file: back up, restore and import. One caveat, one direction: personas built on the hosted app carry more than the Open creator holds, so they arrive simplified on your own instance.

When this doesn't work

Sync is the only thing a server refuses:

MessageWhat it meansWhat to do
Sync not availableYou are not subscribed to Pro.Subscribe, or keep working locally; local data is untouched.
Storage quota exceeded. Please upgrade your plan or free up space.You are at 100% of cloud space: uploads stop, reads and deletes still work.Delete synced items, or move to the larger quota.

Missing chats: Where did my chats go?. Devices: Will my chats appear on my other devices?.

© 2026 Token Fabrics·Built with passion in San Diego