Skip to main content

talonic_get_pricing

Returns the credit pricing catalog: fixed per-unit credit rates, their EUR equivalents, the credits-per-EUR conversion rate, and processing-mode multipliers (such as batch at 0.5x). Because the rates are fixed rather than token-based, an agent can read them and compute the exact cost of a planned job before spending a credit. The endpoint is public, so it works even before an API key is provisioned.

When to use

  • Estimating the cost of a planned extraction, structuring, or matching job before running it.
  • Answering a user question about how much Talonic costs per page or per operation.
  • Deciding whether a workload fits a budget, especially before signing up.

When not to use

  • For the workspace's remaining credit balance: use talonic_get_balance.
  • For what the workspace has already spent and on what: use talonic_get_usage.

Response shape

  • currency: billing currency for the EUR equivalents (always EUR).
  • credits_per_eur: conversion rate between credits and EUR (1,000 credits = 1 EUR at the standard rate).
  • multipliers: processing-mode multipliers keyed by mode (e.g. realtime: 1, batch: 0.5).
  • units: array of catalog lines, each with unit, label, credits, eur, and free.
Tool input
{}
Tool response
{
  "currency": "EUR",
  "credits_per_eur": 1000,
  "multipliers": { "realtime": 1, "batch": 0.5 },
  "units": [
    { "unit": "page_ingest", "label": "Document page ingestion", "credits": 100, "eur": 0.1, "free": false },
    { "unit": "structuring_cell", "label": "Structuring cell (LLM-reasoned)", "credits": 20, "eur": 0.02, "free": false },
    { "unit": "registry_resolved_cell", "label": "Structuring cell (Field Registry resolved)", "credits": 0, "eur": 0, "free": true },
    { "unit": "intelligence_op", "label": "Intelligence operation (matching, cases, reconciliation)", "credits": 100, "eur": 0.1, "free": false },
    { "unit": "delivery", "label": "Delivery (webhook / export)", "credits": 0, "eur": 0, "free": true },
    { "unit": "validation", "label": "Validation / quality check", "credits": 0, "eur": 0, "free": true }
  ]
}

Example: quote a workload before running it

Agent estimates the cost of a 20-page batch extraction
// User: "How much will it cost to extract this 20-page contract in batch mode?"
// Agent reads pricing:
// talonic_get_pricing → {}

// page_ingest = 100 credits, batch multiplier = 0.5
// 20 pages * 100 credits * 0.5 = 1000 credits = 1.00 EUR

// Agent: "Extracting 20 pages in batch mode costs about 1,000 credits
//  (1.00 EUR). Realtime would be 2,000 credits (2.00 EUR). Proceed in batch?"

Cost-aware agents call talonic_get_pricing to turn a workload into a concrete quote before committing to it. Multiply the relevant per-unit rate by the expected unit count, then apply the processing-mode multiplier. Because the catalog also returns the credits-per-EUR rate, the agent can present the estimate in either credits or euros without hardcoding any number.

The catalog distinguishes billable units from free ones. Document page ingestion, structuring cells filled by fresh model reasoning, and intelligence operations carry a cost; Field Registry resolved cells, delivery, and validation are free. A document whose fields are answered entirely from the Field Registry therefore incurs only its page-ingestion cost, which an agent can surface as a reason to reuse schemas across similar documents.

Frequently asked questions

Does talonic_get_pricing require an API key?+
No. The pricing catalog is public, so an agent can read rates before signing up or provisioning a key. Call it with no arguments; it is read-only and safe to call at any time.
How does an agent compute the cost of a job from the catalog?+
Multiply the per-unit credits by the number of units (pages, cells, or operations), then apply the processing-mode multiplier (1 for realtime, 0.5 for batch). Divide by credits_per_eur to express the result in euros.
Which operations are free in the catalog?+
Field Registry resolved cells, delivery, and validation cost zero credits. Only fresh ingestion, fresh model-reasoned structuring cells, and intelligence operations (matching, cases, reconciliation) are billable.