> ## Documentation Index
> Fetch the complete documentation index at: https://docs.ariacompute.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Cost by Competence Tier

> Competence-tier (competence tier) attributed cost for Aria Compute ROUTER, plus estimated savings from competence-aware downgrades.

Aggregates requests by `competence_tier` (`small`/`mid`/`large`) and estimates what the cost would have been if every request had gone to the `large` tier. The difference is the savings from competence-aware downgrades.

## Endpoint

```http theme={null}
GET /v1/router/cost/by-competence
```

## Authentication

Authenticate with a session cookie or `Authorization: Bearer <api-key>`.

## Response

**by\_tier** `object`

Per-tier aggregated cost, each tier containing:

**tier fields**

**requests** `integer`

Request count for the tier.

**cost\_usd** `number`

Tier cost in USD.

**tokens** `integer`

Total tokens for the tier (prompt + completion).

**actual\_cost\_usd** `number`

Actual total cost in USD.

**notional\_large\_cost\_usd** `number`

Notional cost had everything gone to the large tier (USD).

**estimated\_savings\_usd** `number`

Estimated savings from competence-aware downgrades (USD), the greater of `notional_large_cost_usd - actual_cost_usd` and `0`.

## Example

```bash theme={null}
curl -s https://router.example.com/v1/router/cost/by-competence \
  -H "Authorization: Bearer <api-key>"
```

```json theme={null}
{
  "by_tier": {
    "small": { "requests": 120, "cost_usd": 0.30, "tokens": 9000 },
    "mid":   { "requests": 80,  "cost_usd": 1.20, "tokens": 24000 },
    "large": { "requests": 10,  "cost_usd": 2.50, "tokens": 5000 }
  },
  "actual_cost_usd": 4.00,
  "notional_large_cost_usd": 16.00,
  "estimated_savings_usd": 12.00
}
```


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.