// Pricing

Simple, transparent pricing

Start for $1.99 your first week, then $20/month (Pro). Auto-renews monthly. Cancel anytime.

First week

Best deal
$1.99/first week

then $20/mo (Pro)

  • All open cloud models
  • Generous monthly usage
  • 1 concurrent deployment
  • OpenAI-compatible API
  • Email support
Get started

Pro

Most popular
$20/mo

  • All open cloud models
  • Generous monthly usage
  • Up to 3 concurrent deployments
  • OpenAI-compatible API
  • Email support
Get started

Max

$100/mo

  • Everything in Pro
  • Highest monthly usage
  • Up to 10 concurrent deployments
  • Priority compute
  • Priority support
Get started

Plans are billed monthly and renew automatically until cancelled. Introductory pricing converts to the standard plan price after the introductory period unless cancelled. See our Terms of Service and Refund Policy for details.

// FAQ

Frequently asked questions

What happens once my first week ends?

Your $1.99 first-week plan rolls over to Pro at $20/month automatically unless you cancel before the week is up. The exact renewal date is spelled out at checkout.

Can I cancel whenever I want?

Yes — cancel from your account at any time. Your plan keeps working through the end of the current billing period, with no further charges after that.

What does each plan include?

All plans can run every open cloud model; what changes is concurrency and usage. First week runs one model at a time with light usage, Pro runs up to three at once with more headroom, and Max runs up to ten with the highest usage.

Do I have to bring my own GPUs?

No. Everything runs on managed, GPU-backed cloud infrastructure — nothing for you to provision, rent, or operate.

Which models can I run?

Every open cloud model listed on our Models page — text, vision, tools, and thinking models — plus new ones as they ship. GLM-5.2 is the current highlight.

Do the models support tool calling and agents?

Yes — we expose the tool-calling features each provider ships, ready for agent workflows; the precise behavior comes down to the model itself.

What model weights do you run?

We run each provider’s officially released weights, with hardware acceleration switched on wherever it’s available.

How is usage measured, and what are the limits?

We measure usage by the compute your runs actually use — mainly GPU time — rather than a fixed token count. Every plan comes with an allowance that resets on a set cycle, and bigger or heavier models draw it down faster. Check your usage anytime in your account.

What if I need more usage?

Once the included allowance is gone, Pro and Max can top up with extra usage, or you can move to a higher plan for more concurrency and headroom. We email you the moment you reach 90% of your allowance.

Is the API compatible with OpenAI?

Yes. Aim your existing OpenAI client at our base URL and leave your code as-is. The Docs have the details.

Do you train on my data?

No — your prompts and outputs are never used to train models. Our Privacy Policy covers exactly how data is handled.

How do refunds work?

Billing periods that have already begun are non-refundable unless the law requires otherwise. Hit a billing error or duplicate charge? Contact support — see our Refund Policy.