Pricing
Pay for what you send.
One rate for every request. Answers, thinking, forecasts, images and generated text are included, so the price never depends on how hard the question was.
Pay as you go · announced launch price
$0.042per 1M input tokens
Output tokens: free
Free during the preview. This is the price paid usage launches at, and the rate every request is metered at today.
- Every answer and its full probability distribution
- Auto-thinking on close calls, however long the thought
- Worked-out forecasts for how-likely questions
- Images in the state, and the text generate writes
- Several questions about the same state in one request
Volume & on-premises
Let's talk
Committed volume, a dedicated deployment, or Wity running inside your own network: same model, same API.
What a decision costs
Input tokens are what you send. Prices below are per 1,000 decisions.
Each question is billed with the state it reads, so three questions about one ticket cost about three times one question. Keep the state to what the decision needs. See billing in the docs.
Questions
What exactly counts as input tokens?
Your request as Wity reads it: the state, the instructions and the option descriptions, counted once per question. The response reports it in usage.input_tokens, so you can reconcile every call.
Is thinking billed?
No. When Wity reasons before answering, or checks an answer from several angles, that work is on us. You pay the same for a thought answer as for a direct one.
Is there a minimum or a subscription?
No. You pay for the tokens you send. Volume pricing and on-premises deployments are available; talk to us.
How do I pay?
The API is in a free preview: keys are issued on request and nothing is charged. Self-serve sign-up and card payment launch in the coming days at the announced price above, $0.042 per million input tokens with output free. Every response already reports usage.input_tokens, so you can see exactly what a workload will cost.