Read the docs
Product preview · In development

A small, explicit inference flow.

Discover a model, review a quote, sign a capped request, and retain a receipt. Paid inference and SDK availability are in development.

  1. {n}

    Pick a stage

    Explore the request lifecycle. These examples are planned API pseudocode, not a working SDK.

  2. {n}

    Choose the exact request

    Select a supported model revision, provide the prompt, and set the maximum output length. The initial catalog is pending validation.

  3. {n}

    Inspect the example

    // Planned API pseudocode
quote = requestQuote(model, prompt, outputLimit)
review(quote.rates, quote.maxCharge, quote.expiresAt)
  4. {n}

    Review the rates and maximum charge

    A quote is planned to expire after 30 seconds. A changed prompt requires a fresh quote. Capacity and funds are not reserved yet.

  5. {n}

    Know the boundary

    Accept only while the quote and capacity are valid. An unexpired quote may still encounter unavailable capacity.

    Read the docs ↗