Jev cost per decision

Enter your own measurements. The worksheet runs only in your browser; it sends nothing and calls no model.

Average usage.input_tokens from your own requests. Required; there is no default.
One decision = one question answered.
The state is billed once per request, however many questions it carries. Your token count above must match this batch size.
Extra requests per 100 requests. Counted as fully billed (worst case; billing of failed requests is not documented).
Decisions you send on to another model or a person (low confidence, errors).
What one decision costs on your current model or process. Needed for the fallback and break-even rows.

Jev list price used: $0.042 per million input tokens; output tokens free (TypeSafe models page, checked 7 Oct 2026, 05:19 UTC). Gateways may bill differently; on OpenRouter use each response's usage.cost.

Result

Enter your measured input tokens per request to see the cost.

Reference only, not a default: Opper reported that Jev adds about 257 input tokens to every request (25 Sep 2026, operator-reported, Reported). TypeSafe's own API examples show 296–318 input tokens for a one-sentence state and one short question.