Skip to main content
Once you’re signed in, you can test a prompt right away. You don’t need an API key to start — your account comes with a credit wallet and access to Viorant’s managed models.

Start with credits

Every account gets 2,500 credits, once. Pick a Viorant tier and run — no card, no key, no setup.

Or bring your own key

Add your OpenAI, Anthropic, Google, or other provider key in the Model Master. Keys live in your OS credential store and never leave your machine — and cost no credits.

​
Run a prompt in three steps

1

Create a prompt in your Space

Add a new prompt to your Space and write a system instruction — e.g. “You are a concise assistant. Answer in one sentence.” It opens in the Prompt editor.
2

Pick a model

In the Test panel, choose a model. Use a Viorant tier — Medium or Low — to run on credits, or add your own provider key and pick one of yours.
3

Test it

Send a message and watch the response stream back live. Switch between single-shot and chat to see how the prompt behaves.
Start on Low while you’re iterating on wording — it’s the cheaper tier and usually enough to tell whether a prompt is heading the right way. Move up to Medium for the runs that matter.

​
Keeping an eye on credits

Your balance sits next to your profile. For the breakdown, open the profile menu and choose Usage — you’ll see a day-by-day history and can expand any date to see the individual runs behind it. If the balance reaches zero, managed models stop and the Hub prompts you to add your own key, which has no credit cost.

​
Where to go next

Understand the building blocks

Agents, the Masters, Skills, Connectors, and Models.

Compose an agent

Combine a prompt, a model, and skills into one runnable agent.