Cost calculator
Frontier open-source models through Dynoyard, versus what the same tokens cost on a frontier US lab. Live prices, cache-aware. Adjust the numbers to your traffic.
Share of input tokens served from cache (repeated system prompts / context).
Dynoyard prices are live from the catalog. Frontier baselines are published list prices for a representative GPT-4-class / Claude-class model and are illustrative, not a Dynoyard product. Estimates only — your real cost depends on your model mix and cache rate. This is not a comparison against other gateways: per model, an aggregator's price can be at, above, or below ours.