Data
LLM pricing API and static JSON export
LLM pricing API and static JSON export explains How to consume ByteCosts LLM pricing data as a static JSON export, with schema notes, update cadence, limitations, and source policy. ByteCosts publishes static AI pricing data for 4,588 priced models across 167 providers, including 310 official price rows. The public exports are generated from committed JSON during the build, so reading the data never calls a provider API and never needs a secret. Each row keeps the provider, model, input and output price where known, unit, currency, context, source URL, last checked date, and confidence field. Use these exports for planning and internal analysis, but verify the provider source before production billing decisions because list prices can change between manual refreshes. Any calculation built from the exports is an estimate, not a final billing total, and real usage can vary with traffic shape, retries, caching, taxes, and discounts.
JSON schema notes - Open the public export →
Public export schema
The JSON and CSV exports intentionally expose only public-safe fields needed for pricing analysis.
| Field | Meaning |
|---|---|
| provider | Provider display name. |
| model | Model name as published by the provider. |
| input | Input price in the row unit and currency, or blank when unknown. |
| output | Output price in the row unit and currency, or blank when unknown. |
| unit | Pricing unit, usually USD per 1M tokens. |
| sourceUrl | Provider page the rate was read from. |
| lastCheckedAt | Date the rate was last verified. |
| confidence | Evidence field for the row. |
Use it for
- Internal AI cost models and procurement comparisons.
- Search and answer-engine ingestion of source-backed pricing rows.
- Spreadsheet analysis of provider, model, and confidence coverage.
- Reproducible citations that include a snapshot date and provider source URL.
Limitations
- Rows are list prices and exclude negotiated discounts, committed-use discounts, taxes, and enterprise terms.
- Unknown prices stay blank rather than being inferred.
- The public export is static and refreshes only after the repository data is manually refreshed and committed.
- Calculations built on these rows are planning estimates, not final billing totals, and real usage can differ from modeled assumptions.
- Provider source pages remain authoritative for live billing decisions.
Limitations before production billing decisions
Treat ByteCosts calculations as planning estimates, not final billing totals. Real invoices can differ because token mix, retry rate, cache hit rate, rate limits, taxes, gateway fees, regional pricing, and negotiated discounts change the effective cost.
Verify the provider source before production billing decisions, then compare the estimate with your own logs or invoice once production traffic is live.
Frequently asked questions
Is this a live API?
No. ByteCosts serves static JSON, CSV, and markdown files from the public site. They behave like API resources for clients, but the site has no server route and does not refresh data on request.
How often is the dataset updated?
The snapshot shown here was generated at 2026-07-19T21:42:52.054Z. Refreshes are manual and local, then committed before publication. Build and preview commands do not fetch provider data.
Can I cite the dataset?
Yes. Cite the ByteCosts URL, the snapshot date, and the provider source URL for any price you quote. Treat calculator output as an estimate, not an official provider quote. Verify the provider source before production billing decisions.
LLM pricing API and static JSON export. ByteCosts. Updated July 19, 2026. https://bytecosts.com/data/llm-pricing-api/