Back to products & servicesRUNLUME / RELAY

AI Token Relay Service

Support AI applications with model API access, request forwarding, and usage integration, covering interface compatibility, credential isolation, limits, and troubleshooting. Confirm upstream model sources, billing, data flows, and scope in the service plan. This is distinct from an AI Token Pack.

Product and service scope

  • Model API compatibility and request forwarding
  • Credentials, authorization scope, and allowance configuration
  • Call records, usage, and billing review
  • Upstream permissions, data handling, and log retention review
  • Agree on response arrangements, models, and service scope

Industries: Software & internet

Token relayModel APIsAPI integration

These are product capabilities. Your selected plan or service agreement determines delivery scope, available features, and resource allowances.

Explore capabilities and workflows

Before you purchase

  1. Business goals and service scope
  2. Deliverables, timeline, and acceptance criteria
  3. Quotation, payment arrangements, and support
COMMON QUESTIONS

API relay: routing and usage boundaries.

How does API token relay differ from buying a token pack?

Relay services concern interface adaptation, routing, credential isolation, and usage management. A token pack supplies usage resources within its defined scope. Agree whether they are combined, which upstream providers are used, and how usage is billed. Relay access does not imply unlimited model resources.

Does API compatibility mean every model and parameter is supported?

No. Check endpoints, parameters, context limits, and usage restrictions for each upstream provider and model, then test the client. Compatible formats do not establish identical behavior, output, or billing. The confirmed integration scope defines support.

How are relay data flows and log retention defined?

Before use, identify relay nodes, upstream services, transmitted fields, access controls, and log retention, along with data-processing and billing responsibilities. Limit sensitive inputs to authorized purposes. Distinguish model usage, service fees, and metering rules for failed or exceptional calls.