Calculate costs and private prices

Turn a deployment and a realistic token profile into a deterministic quote before cost reaches production.

What the cost engine prevents

Provider prices use different units and can include cached-input or batch discounts. A deterministic ALLM quote lets product, finance, and engineering compare the same usage profile without spreadsheet drift.

Calculate a quote

1. Copy a deployment public ID from Registry.
2. Enter representative input, cached-input, and output token counts.
3. Enable Batch request only if the workload will actually use batch execution.
4. Select Calculate quote. The latest result shows the total and line-item calculation returned by the live API.

Interpret the result

total_usd

The deterministic total for this usage profile.

line_items

The contribution from input, cached input, output, and applicable adjustments.

rate_card

The rates and evidence timestamp used in the calculation.

deployment_id

The exact serving deployment priced—not only the model release.

Private price books

Pro and Business workspaces can store negotiated rates, regional adjustments, batch discounts, or commercial uplift. Rules match a deployment, provider, model, or region and are versioned as workspace data.

[
  {
    "match": { "provider_id": "openai", "region": "eu" },
    "rate": {
      "multiplier": 0.90,
      "region_multiplier": 1.05,
      "batch_multiplier": 0.50,
      "uplift_multiplier": 1.02
    }
  }
]

Create a new price book when contract terms change rather than silently editing the assumptions behind an earlier decision.

Limits and access

• Public quotes are available without a paid price-book feature.
• Private price books require Pro or Business and an owner, admin, or billing role.
• Use integer token counts and a representative workload profile.