Your AI subscription and models
On the Standard plan the agent runs on a subscription that belongs to you. That is why it costs $350 per month instead of the price of the model usage on top.
Why my subscription and not yours?
Two reasons, both in your favour. You pay the model provider directly at their price, with no margin from us. And your usage limit is yours alone — it is not shared with other customers, so a busy day for someone else never slows you down.
What runs the agent's reasoning?
An AI subscription that belongs to you. We recommend Anthropic Claude (a Max subscription). Max 20x works best; Max 5x also works, but its limit runs out four times sooner. You can also run the agent on ChatGPT (a Pro subscription), from the start or added later. The subscription is issued in your name and stays yours: we do not resell it and it is not bundled into the plan; on the Assisted plan we help you connect it.
Which subscription should I buy?
Buy the plan of the AI provider that gives the largest usage allowance for the money you are willing to spend — that is what your agent's stamina comes down to. Pay-as-you-go API credits are dramatically more expensive per unit of work than a flat subscription, so a subscription is the sane default.
How do I connect it?
In the dashboard, open the agent's page, press "Change" in the "Settings for all chats" card, then Add Claude subscription. A one-time secure link to Anthropic opens: you sign in there under the account you want, copy the confirmation code and paste it back into the dashboard. The password is never typed into a chat and we never see it.

On the second step, under the code field, there is an Apply subscription to agents list. One pass can hand the same subscription to several of your agents — tick everyone who needs it.
The browser remembers the previous sign-in and the link will take you into the same Anthropic account. The dashboard then tells you the key was refreshed and no new subscription was added.
Subscriptions belong to the agent, not to the account
A subscription is connected to one specific agent. With several agents, each has its own list and its own active account: connecting a subscription for one does not connect it for the rest. You can hand it to the others from the same Apply subscription to agents list while connecting.
If a subscription already works for one of your agents, you can give it to another one later, with no link and no code. In My agents press Subscriptions: it shows which subscription each agent, chat and topic runs on. Next to the agent press Move and pick the subscription from the list, and the agent switches to it by itself in about a minute, keeping its memory. An agent without its own subscription yet sits in the EmAGI subscription strip with a Move ... to your own button. A subscription that no agent has yet is connected with + Connect a subscription on the same page or on the agent's page.
Can I connect two?
Yes, and it is the single most useful setting on the card. The order of the list is the queue: the top subscription is the working one, the next is the standby.

What each control does:
- Make active — move the agent onto a subscription that is already connected. No link and no code needed: the agent restarts in about a minute and carries on from where it was.
- Up arrow — move a subscription up the queue, i.e. make it the first standby.
- Cross — remove the subscription from this agent.
- Switch to the standby when the limit is hit — on hitting the limit the agent moves to the next one in the queue by itself instead of waiting for a reset.
- Return to the primary when the limit clears — once the primary's window resets, the agent goes back to it on its own.
The limit belongs to the Anthropic account, not to the connection. If both connections lead to the same account they share one quota — the dashboard says exactly that: not a standby. A standby only makes sense on a separate account.
A second subscription matters most because of the weekly window: the five-hour window clears by itself the same day, the weekly one only on its own reset day. How the windows work: subscription limits and spend.
The switch shows an error. What now?
The most common cause is an expired link — each link works once and for a short time. Click "Get link" again (or ask the agent for a fresh one) and complete it in one go. If a fresh link does not work either, write to support@emagi.ai and we will sort it out.
Can I make one chat cheaper, or one chat smarter?
Yes, and the model is set per chat, not per agent. If a chat is mostly short confirmations and lookups, a lighter model handles it at a fraction of the cost. If a chat is architecture, code or long multi-step work, the strongest model earns its keep there.
The simplest way is to tell the agent in that chat: "work on a cheaper model here" or "this chat is heavy, use the strongest one". It switches itself, and its memory of the chat is untouched. Or pick the model yourself in the dashboard: on the agent's page, in "Chats & thought streams", in that chat's row. A good agent will suggest the switch itself once it sees that a chat is consistently simple.
What about images, video, voice and other tasks?
For generating pictures, video, music, speech and 3D, and for specialised jobs like transcription, the agent does not use its reasoning model at all — it calls a hub of purpose-built models through a single gateway and picks the right one per task. That allowance is part of your plan; the VIP plan carries a five times larger one.
What about Gemini, or another provider?
The agent runs on a Claude or ChatGPT subscription. As experimental options you can connect your own API key from DeepSeek, Venice or Hugging Face; honestly, in agent work ChatGPT and DeepSeek reason a little weaker than Claude. On any of them the employee stays the same, with its memory, skills and access, and you choose what it runs on, down to a single chat, on the Subscriptions page. If another provider matters for your work, write to support@emagi.ai and say which one and why; that is how we decide what to build next.
Do I buy tokens from you?
No. The reasoning runs on your own subscription, in your name. What you pay us for is the agent itself: its isolated environment, its memory, its skills and the platform around it. See what you pay for.