Usage

Track token consumption and cost attribution across projects, teams, and features. Eridian records usage at inference time so FinOps and platform teams can reconcile invoices against granular request metadata without estimating from provider bills alone.

Get Usage

GET /v1/usage?project_id=prj_legal_001&start=2025-06-01&end=2025-06-30
Authorization: Bearer eridian_sk_...

Required scope: usage:read

Query Parameters

Example Response

{
  "project_id": "prj_legal_001",
  "period": { "start": "2025-06-01", "end": "2025-06-30" },
  "total_tokens": 2847291,
  "total_cost_usd": 4823.17,
  "by_model": {
    "gpt": { "tokens": 1928471, "cost_usd": 3210.44 },
    "gemini": { "tokens": 918820, "cost_usd": 1612.73 }
  },
  "by_feature": {
    "semantic_cache": { "tokens_saved": 847291, "cost_saved_usd": 1204.33 },
    "rag": { "tokens": 412847, "cost_usd": 892.11 },
    "pii_redaction": { "tokens": 2847291, "cost_usd": 0 }
  }
}

Per-Request Attribution

Every inference response includes eridian.cost_usd and is attributed to:

• Project and API key
• Routing policy and selected model
• Feature flags (cache, RAG, PII, structured output)
• Template ID and version (if used)

Export usage data via the Observability dashboard, Usage API, or webhooks (budget.threshold, usage.daily_summary).

Included Tokens and Overage

Commercial capacity is defined in your order form. The usage API reports consumption for governance and attribution; it does not expose the commercial terms of your agreement. See the pricing page for plan details.

Webhook Payload

{
  "event": "usage.daily_summary",
  "project_id": "prj_legal_001",
  "date": "2025-06-15",
  "total_tokens": 98472,
  "total_cost_usd": 167.32
}

See Governance for budget configuration, Pagination for starting_after on large exports, and Authentication for API key scoping.

Python

usage = client.usage.retrieve(
    project_id="prj_legal_001",
    start="2026-09-01",
    end="2026-09-30",
    group_by="model",
)
print(usage.total_cost_usd, usage.by_model["gpt"])

TypeScript

const usage = await client.usage.retrieve({
  projectId: "prj_legal_001",
  start: "2026-09-01",
  end: "2026-09-30",
  groupBy: "model",
});