AI inference is priced by input and output tokens. Managed compute is individually quoted. Both use the same account-specific USD service credit.
Rates are shown in USD per one million tokens, with a charge record for each completed request.
Published rates and available models appear here when the inference service is activated. No API usage is charged while a model is unavailable.
Create a workspaceAdd credit using an activated payment method in the console. Your credit covers AI usage and accepted managed compute services.
Compute is scoped around your actual workload. Review a written quote for resources, region, duration and support scope before accepting.
The platform reserves an estimated maximum charge before sending the request upstream. A successful response is charged using actual returned input and output token counts at the published model rates. Any unused reservation is released.
A request that cannot be covered by available credit is declined before it is sent to the upstream model. Monthly API-key spend limits can also stop further use.
No. Service credit is specific to your RDGlobal account and is used only for eligible services. Refund requests for unused purchased credit are reviewed under the Refund Policy.
Managed compute is quoted individually. The accepted quote defines the resources, service period and support scope. Submit a request for a proposal.