Compute tokens: prepaid LLM inference as a service

We align inference tokens, LLM API credits, and AI compute into a single purchasable, redeemable, reconcilable balance—closer to cloud prepayment than to tradable crypto assets.

Product boundary (planned)

Wallet tokens redeem only for inference and related services on-platform; closed-loop redemption under published terms. Model-version and region multipliers can reflect different burn rates for finance and chargeback.

Capabilities behind the keywords

  • Inference metering: per-request logs—model, region, latency, billing dimensions—with CSV export and finance hooks.
  • Budget & kill switch: budgets at org, project, and key scope; automatic trip on anomalous traffic to reduce key theft and misconfiguration risk.
  • Multi-model routing: one balance pool can route across providers (implementation evolves) while customers see one bill.
  • Global SKU: market-specific packages, tax-inclusive display, and availability rules aligned with privacy and residency commitments.

Why a “.store” domain

buytokens.store emphasizes buy, a token unit of account, and a store front door. At launch, the domain can host the global top-up, invoice, and usage portal, while technical docs live on a dedicated subdomain.