specification: FinOps Framework
specificationVersion: '1.0'
schema: https://www.finops.org/framework/
alignedWith:
  framework: FinOps Foundation Framework
  frameworkUrl: https://www.finops.org/framework/
  dataSpec: FOCUS
  dataSpecVersion: '1.3'
  dataSpecUrl: https://focus.finops.org/focus-specification/v1-3/
provider: Qubrid AI
providerId: qubrid-ai
publisherName: Qubrid AI
serviceCategory: AI Infrastructure
created: '2026-05-04'
# Provenance stamped 2026-08-11: this artifact was written by the API Evangelist
# bulk sweep dated 2026-05-04, not harvested from the provider. See roadmap#35.
method: generated
modified: '2026-05-05'
reconciled: true
tags:
  - Artificial Intelligence
  - GPU
  - Inference
  - FinOps
  - FOCUS
description: FOCUS-aligned FinOps for Qubrid AI - per-token serverless inference plus GPU virtual machine
  rentals on multiple commitment terms and dedicated NVIDIA HGX bare-metal annual contracts.
sources:
  - https://www.qubrid.com/pricing
  - https://www.qubrid.com/serverless-inferencing
billingModel:
  pricingCategory: Pay-As-You-Go + Committed Use
  billingFrequency: On-Demand
  billingCurrency: USD
  chargeCategories:
    - Usage
    - Purchase
    - Tax
    - Credit
    - Adjustment
focusColumns:
  ServiceName: Qubrid AI
  ServiceCategory: AI Infrastructure
  ProviderName: Qubrid AI
  PublisherName: Qubrid AI
  InvoiceIssuerName: Qubrid AI
  BillingCurrency: USD
  ChargeCategory: Usage
meters:
  - name: tokens_input
    description: Input tokens processed by the OpenAI-compatible serverless inference endpoint, billed
      at per-model rates set in the Qubrid platform.
    unit: token
    aggregation: sum
    dimensions:
      - model
      - api_key
  - name: tokens_output
    description: Output tokens generated by the OpenAI-compatible inference endpoint, billed at per-model
      rates.
    unit: token
    aggregation: sum
    dimensions:
      - model
      - api_key
  - name: gpu_hours
    description: GPU virtual machine hours consumed under On-Demand or 1 Week / 1 Month / 3 Month / 1
      Year reservation terms.
    unit: gpu-hour
    aggregation: sum
    dimensions:
      - gpu_model
      - reservation_term
      - region
  - name: bare_metal_subscription
    description: Annual NVIDIA HGX 8-GPU bare-metal contract; published rates from $89,500/year (A100)
      to $380,000/year (B300).
    unit: year
    aggregation: sum
    dimensions:
      - gpu_model
      - contract_term
  - name: prepaid_credits
    description: Account credits funded by deposit; new accounts receive a $1.00 free credit on a $5
      first deposit.
    unit: credit
    aggregation: sum
    dimensions:
      - account
principles:
  - name: Visibility
    description: Track inference token consumption per API key and per model in the Qubrid platform
      console; review GPU VM utilization to align reservation term with steady-state demand.
  - name: Allocation
    description: Issue separate API keys per consuming team or product so token spend can be attributed;
      tag GPU VM and bare-metal allocations against the workload owner.
  - name: Optimization
    description: Move steady-state GPU workloads from On-Demand to 1 Month, 3 Month, or 1 Year reservations
      for committed-use pricing; choose annual bare-metal HGX contracts (A100/H100/H200/B200/B300) when
      utilization justifies the fixed cost; right-size GPU model to match the deployed model size.
  - name: Accountability
    description: Assign a budget owner per API key and per dedicated GPU/bare-metal contract; track
      bare-metal annual commitments separately on the FinOps balance sheet because they are large fixed
      costs that should not be confused with PAYG inference spend.
maintainers:
  - FN: Kin Lane
    email: kin@apievangelist.com