No free snack is open right now.
When the next one opens, its claim terms will appear here.
Choose leading AI models through one OpenAI-compatible API. Check live input, output, and cache token prices before sending a request.
Free tokens and paid model packs live here. Each paid pack is reserved for the exact model on its card.
Enjoy a small token snack for the free models that are available today.
When the next one opens, its claim terms will appear here.
Use a paid pack only with its named model. It never silently falls back to PAYG.
A pack appears here only after its model and price are published for sale.
Prices are shown per one million input, output, or cache tokens. Model availability is refreshed every 30 seconds.
claude-fable-5ClaudeRpĀ 71.850$3.9566Official claude-fable-5.1ClaudeRpĀ 72.500$3.9924Official claude-haiku-4.5ClaudeRpĀ 7.250$0.3992Official claude-opus-4.6ClaudeRpĀ 36.250$1.9962Official claude-opus-4.6-thinkingClaudeRpĀ 36.250$1.9962Official claude-opus-4.7ClaudeRpĀ 36.250$1.9962Official claude-opus-4.8ClaudeRpĀ 36.250$1.9962Official claude-opus-5ClaudeRpĀ 36.250$1.9962Official claude-opus-5.5ClaudeRpĀ 29.000$1.597Official claude-sonnet-4.5ClaudeRpĀ 21.750$1.1977Official Every model has separate live rates for input, output, and cache tokens. The model price is confirmed before the request runs.
Enter a realistic token count. This estimate reads the same public price data as the model catalog above.
Add credit through EidosStack checkout. Your balance becomes available after payment is confirmed.
For a paid model with an active package, the package quota is used first. If there is no active package for that model, the request uses your PAYG wallet. An exhausted active package returns an insufficient-package-quota error and does not silently charge PAYG.
Prompt text and completion text are not shown in usage history or placed in cache keys. Eligible requests can use a short-lived response cache that is isolated to the same account.
The live catalog shows the models that are currently available, their token prices, and their cache prices. It refreshes availability every 30 seconds.
See endpoints, authentication, limits, and request examples.
Use input, output, cache, and a realistic request scenario before adding credit.
Learn why input and output tokens are priced separately.
Check live availability, token units, integration, and subscription boundaries.
Choose a billing approach based on your workload.
See how request content, usage records, and service providers are handled.
Use maintained examples that start from the live model catalog and keep credentials in environment variables.