Clear prices for each model.
Available catalog models show a default input and output rate in USD per 1,000,000 tokens.
Current rates
A missing price means the model is not currently callable. Check availability before sending requests.
How a request is charged
Before a request is sent, Infera reserves an estimate based on the prompt, output limit and current model rate. When the request finishes, the unused reservation is released and actual usage is charged.
If a request produces no output, no inference charge is made.
Usage source
Receipts distinguish reported token counts from estimated usage when counts are unavailable.
Funding
Your prepaid Infera balance is used for requests. Current funding options and availability are shown in your dashboard.