All plans

Nighttime Pass

What's included

  • One night

    Applies to the current time window if you book during it, otherwise to the next one – and ends automatically afterwards.

  • Unlimited tokens

    No quota – you can process as much text as you like.

  • Up to 16 parallel requests

    Further requests are rejected until a running one has finished.

  • Only 22:00–06:00

    Outside this time window (German time), the plan doesn't accept requests.

  • 1 API key

    For your tools and scripts. The web chat in the portal doesn't need its own key.

  • IP binding: 10 min

    Within 10 minutes, a key can only be used from one IP address – protection against sharing and leaked keys. You can reset it yourself once a day.

  • Context window up to 1,000,000 tokens

    This is how much history, files and instructions a single request may contain. Longer requests are rejected – then shorten the history or start over.

After booking

  1. 1Start right away in the chat – no further setup.
  2. 2Or create a key under “API keys”.
  3. 3Enter the base URL and key in your tool or SDK.

Frequently asked questions

What happens when the pass expires?

It ends automatically, with no further costs. Your chats and API keys are kept.

Can I book several passes?

Yes, one after another: if you book a pass while another is still running, the new one starts as soon as the current one ends; the Free Pass doesn't count here. A pass takes precedence over a subscription.

How does IP binding work?

After a request, an API key can only be used from the same IP address for 10 minutes (with IPv6: from the same connection, i.e. the /64 network). During this time, the API rejects requests from other addresses with HTTP 403. If you switch networks – for example to a VPN, to mobile data or between IPv4 and IPv6 – wait a moment or release the binding under “API keys” in the portal; you can do this once every 24 hours per key. This protects you against your access being passed on and against misuse of keys that were published by accident.

What is a token?

Language models split text into tokens – word pieces of about ¾ of a word. Tokens are unlimited in this plan.