Ship telemetry fast, keep attribution stable, and move from ingest to governance without changing your app network path.
Start with direct ingest, validate one stable payload shape, then harden budgets, retries, and incident workflows.
Implementation rhythm
Use quickstart for initial payloads, then operations docs to harden production workflows.
Before sending ingest payload, always fetch GET /v1/public/pricing-models and use the provider/model strings exactly as returned.
Rule: 1) Pull catalog, 2) copy provider/model exactly, 3) send payload.
If provider/model is not found in catalog, ingest falls back to unknown and costUsd = 0 until pricing is added.
| Model | Input Price | Output Price | Effective from | Action |
|---|---|---|---|---|
gpt-4o-mini | $0.15 | $0.6 | 2026-02-06 | |
gpt-4-1 | $2 | $8 | 2026-02-06 | |
gpt-4o | $2.5 | $10 | 2026-02-06 | |
gpt-4-1-mini | $0.4 | $1.6 | 2026-02-06 | |
ada | $0.4 | $0.4 | 2026-02-06 | |
ada-batch | $0.2 | $0.2 | 2026-02-06 | |
babbage | $0.5 | $0.5 | 2026-02-06 | |
babbage-batch | $0.25 | $0.25 | 2026-02-06 |
Use pricing overrides only when contract-specific rates must differ from the default catalog. Keep overrides versioned by effective date and preserve immutable historical snapshots for reporting integrity.
See full guidance: Pricing table overrides: enterprise workflows & auditability.