The models class prices the output of AI labs the way the GPU class prices compute. The Ornn Token Price Index (OTPI) settles a USD-per-million-token value per lab, weighted by transacted token volume, with the Anthropic and OpenAI indices listed on Bloomberg as ORNNTOKA and ORNNTOKO; beside it sit the Token Volume Index, the Workload Cost Index, the Model Frontier dataset and the open-weight versus proprietary split of the model market.
Which token prices are free to read?
Four labs are public: Anthropic, OpenAI, Google and DeepSeek, with the latest settled value and a three-month daily history through the keyless REST API (and through the MCP server with a free Ornn account). Anthropic (ORNNTOKA) and OpenAI (ORNNTOKO) also have market pages with 1/7/30-day changes. The other tracked labs, full history, in-app exports, token volume and workload cost are part of a subscription.
What is the Ornn Token Price Index?
OTPI is a settled USD-per-million-token benchmark per lab, weighted by transacted token volume across the models and tiers that buyers actually used, so it moves when usage shifts between models even if list prices do not. The construction is documented at data.ornn.com/docs/token-price-index.
How does token cost relate to GPU rental prices?
Tokens are produced on rented GPUs, so the OCPI records the input cost of inference and OTPI the output price. The Workload Cost Index prices the request rather than the token: the settled cost per request at the closed labs it covers, with a like-for-like index and tokens per request.
Can I pull token prices from the API?
Yes. The REST API and MCP server expose the latest OTPI value by lab, the list of tracked labs, the Token Volume Index by day and the Workload Cost Index by provider; the free tier covers four labs over the trailing three months and the endpoints are documented at data.ornn.com/docs/token-price-index.