UltraGPT
How can we help?
Self-serve answers for the things people ask most. If yours isn't here, the team replies to email within one business day.
In-app search is faster once you're signed in — this page collects the highlights.
Six places to look
Getting started
Signing in, the interface, running your first chats side by side.
Account & login
Email changes, password resets, devices, deleting your account.
Billing & plans
The flat price, invoices, upgrades, cancellations and refunds.
Models & limits
Which models are included, the 4-hour window, and what unlimited means.
AI tools
The 2,000+ task tools, how they're updated, and requesting new ones.
Privacy & security
Device encryption, what providers see, data retention and exports.
The answers people open most
What exactly is included for $7.90/month?
The Plus plan — chat, image generation, file uploads, internet search and file creation across OpenAI, Anthropic, Google, xAI and more. 60+ models and 2,000+ AI tools are in the workspace; higher tiers unlock exclusive Pro models, Deep Research, voice and a larger message allowance. No credits, no per-model charges.
What does the 4-hour window mean?
Each plan has a message allowance that fully refreshes on a rolling 4-hour window instead of a credit balance to top up. In normal use you won't hit it; if you do, it clears itself within a few hours with nothing to buy.
How do I cancel, and do I lose access immediately?
Account settings → Subscription → Cancel. You keep everything until the end of the period you already paid for, then it stops — no surprise renewal.
Can I get a refund?
Yes — email support within 14 days of the charge and we'll return it. No form to fill in.
Are my conversations private?
Messages are encrypted on your device before they leave it. We relay the sealed payload to the official provider API and don't store chat content in our database. Each provider still applies its own handling on its side, which is part of how we choose them.
A model gave an error or stopped mid-answer. Why?
Usually a provider-side hiccup or a very long input. Retry, shorten the input, or switch to another model for the same thread. If it keeps happening, send us the conversation link.
Is there an API and a CLI?
Both. One endpoint serves every model with the same request shape, and the ultragpt CLI brings the workspace to your terminal. See the docs at /docs.