Models, pricing and limits

The model, what a request costs, the free decisions, and the request limits.

Model

Output tokens are free. A Direct mode decision is billed on the input tokens of all its stakeholders' runs together, and a typed-question request on its input tokens, at the same price. A decision's cost_usd says what it cost.

By default, Direct mode decisions and typed-question requests ask each choice in up to 3 orderings of its options and average the answers, so the answer does not lean on which option is listed first. Averaged requests are billed once: on your input, as without averaging. Their usage shows that input, and meta.orderings how many orderings were asked. A choice whose text names one of its options (for example "Option A is the status quo") is asked once.

Default mode

Billed per input token at $2.50 per 1M, summed over every model call of the request. Measured on real requests:

These figures are for short texts. Every model call reads the whole context, all the options and the stakeholder, so the cost grows with their length, and there is no fixed maximum price. For example, a 19k-character context with 8 long options read about 5.4k input tokens per call, about $6.08 at 450 calls. With every text at its length limit, a 450-call request can cost about $13. A decision's cost_usd says what it cost.

Billing

Eevo is prepaid. Your first account starts with 5 Default mode decisions, each revisable up to 10 times, and 50 Direct mode decisions, free. Free decisions are used before your balance, and a free decision answers with cost_usd 0. Buy credit on Billing in the console, optionally with auto-recharge. When the free decisions and the balance have run out, requests answer 402 insufficient_credit. Failed requests are never charged and use no free decisions.

Requests paid from credit go first on the model. A request on free decisions that finds it busy answers 503 free_capacity_busy: retry after the Retry-After seconds.

Request limits