Pricing
Two plans. Credits on top.
The plan keeps your agent online. Model usage is paid from a prepaid dollar wallet — with a hard spend cap, so there's never a surprise bill. GST/VAT/US sales tax is calculated at checkout.
Starter
standard model · 2 GB
$29/mo
- ▪1 always-on agent
- ▪2 GB memory + skills volume
- ▪Standard reasoning model
- ▪Telegram channel
- ▪Dollar wallet + usage tracking
- ▪$5 included credits to start
Pro
smart model · 4 GB
$49/mo
- ▪1 always-on agent
- ▪4 GB memory + skills volume
- ▪Smart reasoning model
- ▪Telegram channel
- ▪Priority provisioning
- ▪Dollar wallet + usage tracking
- ▪$15 included credits to start
Top-ups
Add credits any time
Buy prepaid credits in one click from your dashboard, any time your balance runs low. We email you before you run out so your agent never goes quiet unexpectedly.
How is this different from ChatGPT?
ChatGPT forgets you between chats and lives in a tab. AgntOS is your own assistant on its own private machine — it remembers your clients, preferences and voice, lives where you already are (web + Telegram), and runs on a dollar budget you set. It gets more useful the more you use it.
What is a Hermes agent?
Hermes is the self-hosted autonomous agent by Nous Research. It reasons, calls tools, keeps long-term memory, and writes its own reusable skills. AgntOS hosts and manages one for you.
Will it run up a huge bill?
No. Your wallet has a hard cap enforced upstream — when credits run out, the agent simply stops until you top up. You only ever spend what you've loaded.
Do I need to know about models?
Never. Pick Standard or Smart and we handle the rest — including the eight background models Hermes uses for side tasks, tuned for cost.
Do I need my own API key?
No — every plan includes managed models with a hard spend cap, so there's nothing to configure. (Bring-your-own-key is on our roadmap.)
Ready in two minutes.
Create an account, shape your agent, connect Telegram, hit launch.
Get your agent →