> The full index of these docs is at [llms.txt](https://u22a8.ai/llms.txt).

# Pricing

What A8 costs per million input tokens, what counts as a token in a verdict, and how billing works.

A8 is metered per token. Each request is billed for the input tokens it reads, as the `usage` on its response reports them.

## 1 Prices

| Model                                   | Input ($ / 1M tokens) | Output ($ / 1M tokens)              |
| --------------------------------------- | --------------------- | ----------------------------------- |
| `a8-1`, your organization’s model       | TBD (price)           | n/a: the output count is always `0` |
| `a8-1-1@reference`, the reference model | TBD (price)           | n/a: the output count is always `0` |

*Table 1.* Prices per model; a pinned version is billed at the price of its model.

- **Input ($ / 1M tokens):** the price of one million input tokens, summed over an organization’s requests in a billing period.
- **Output ($ / 1M tokens):** A8 scores the verdict, so every response reports `0` output tokens.

| Feature               | Price                     |
| --------------------- | ------------------------- |
| Teach                 | TBD (price per teach)     |
| Retract               | TBD (price per retract)   |
| Fitting a new version | TBD (price per fit)       |
| Precedents and recall | TBD (price per 1M tokens) |

*Table 2.* Prices of the features beside judging.

## 2 What counts as a token

- **What A8 reads.** A verdict counts the texts A8 judges with: the criterion, the subject and its standards, and each property’s own criterion. A recall counts the criterion it asks about. A request answered with a `no_question` abstention counts the text of each of its messages.
- **The count on the response.** `usage.prompt_tokens` on Chat Completions and `usage.input_tokens` on Responses and Anthropic Messages report the count billed.
- **A8’s own tokenizer.** A8 counts tokens its own way, so the count for a given text differs from another provider’s count for the same text.
- **More than one choice.** With `n` above `1` on Chat Completions, each choice is computed on its own, and the input counts once per choice.
- **An abstention.** An abstention reads the same input as an answer. Its billing is TBD (abstention billing).

## 3 A worked example

A regression suite judges 10,000 replies against one criterion, with a verdict shape. Each reply runs about 300 tokens and the criterion with its shape about 100, so each request reads about 400 input tokens:

| Item                     | Amount                            |
| ------------------------ | --------------------------------- |
| Requests                 | 10,000                            |
| Input tokens per request | About 400                         |
| Input tokens in all      | About 4,000,000                   |
| Cost at the `a8-1` price | TBD (4 × the price per 1M tokens) |

*Table 3.* A worked example with illustrative token counts; your responses report the real ones.

## 4 Billing

Usage is metered per organization and invoiced TBD (billing period and payment methods). The terms that govern billing are in the [Terms of service](https://u22a8.ai/terms).

## Next steps

- [Rate limits](https://u22a8.ai/docs/rate-limits.md): How A8 limits requests and tokens per organization, the headers that report the limits, and how to raise them.
- [API overview](https://u22a8.ai/docs/api-overview.md): What every request to A8 shares, from the base URL and authentication to the request parts and the response fields.
- [Teaching](https://u22a8.ai/docs/a8-teaching.md): How to teach and retract through MCP, when a teach lands in your organization’s model and what it changes.