aid: parasail specVersion: '0.1' name: Parasail FinOps Surface url: https://www.saas.parasail.io/ description: | Parasail's billing surface is a hybrid of token-metered serverless and GPU-hour reserved capacity. Tokenmaxxing customers can mix Serverless, Dedicated Serverless, Dedicated, and Batch on the same account, each contributing line items to monthly invoices. Aligned to the FinOps Framework / FOCUS data spec dimensions where applicable. billing: unitsOfBilling: - id: input-token label: Input Tokens (1M) surface: Serverless / Dedicated Serverless / Batch description: Pay-per-token input billing, per model. - id: output-token label: Output Tokens (1M) surface: Serverless / Dedicated Serverless / Batch description: Pay-per-token output billing, per model. - id: cached-token label: Cached Tokens (Batch) surface: Batch description: Additional 30% discount on cached tokens within a batch job. - id: gpu-hour label: GPU Hour surface: Dedicated description: GPU-hour billing for reserved capacity, per device SKU (H100, A100, H200, etc.). discounts: - id: batch-discount label: Batch discount amount: 50% off serverless rates - id: cached-discount label: Cached-token discount (Batch) amount: Additional 30% off cached tokens - id: pause label: Pause to stop GPU-hour billing amount: Dedicated deployments billed only while running freeCredits: description: New users receive starter credits for serverless usage. focusMapping: ChargeCategory: - Usage - Purchase ChargeSubcategory: - Tokens - GPU Hours PricingCategory: - Pay-per-token (Serverless / Dedicated Serverless / Batch) - Reserved (Dedicated) CommitmentDiscountCategory: - None by default; volume discounts available via Enterprise contracts. reporting: available: - Per-API-key usage in the Parasail SaaS dashboard - Per-deployment billing for Dedicated capacity - Per-batch cost and token counts in the Batch API response notAvailable: - Programmatic FinOps export API (as of 2026-05)