specification: FinOps Framework
specificationVersion: '1.0'
schema: https://www.finops.org/framework/
provider: CentML
providerId: centml
created: '2026-06-21'
modified: '2026-06-21'
reconciled: false
tags:
- AI
- LLM
- Inference
- Serverless
- GPU
- FinOps
- Cost Management
- FOCUS
description: >-
  FinOps view of CentML spend. CentML bills on a credit-based model where 1
  credit equals 1 USD. Serverless inference is metered by input and output
  tokens with per-token rates that vary by model. Dedicated deployments are
  metered by GPU hardware type and duration on a per-minute basis (per-GPU-hour
  equivalent). Exact per-model and per-GPU-hour rates are not reconciled here.
notes: >-
  Per-model serverless token rates and per-GPU-hour dedicated rates change over
  time; verify against the CentML pricing page during reconciliation.
sources:
- https://centml.ai/pricing/
- https://docs.centml.ai/apps/serverless
- https://focus.finops.org/focus-specification/v1-3/
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/
publisherName: CentML
serviceCategory: AI and Machine Learning
billingModel:
  pricingCategory: Usage-Based
  billingFrequency: Monthly
  billingCurrency: USD
  chargeCategories:
  - Usage
  - Purchase
  - Adjustment
focusColumns:
  ServiceName: CentML Platform
  ServiceCategory: AI and Machine Learning
  ProviderName: CentML
  PublisherName: CentML
  InvoiceIssuerName: CentML
  BillingCurrency: USD
  ChargeCategory: Usage
  PricingCategory: Usage-Based
meters:
- name: serverless_input_tokens
  description: Input tokens sent to serverless chat / text completions, billed per 1M tokens per model.
  unit: tokens
  aggregation: sum
  dimensions:
  - account
  - model
  - api
- name: serverless_output_tokens
  description: Output tokens generated by serverless endpoints, billed per 1M tokens per model.
  unit: tokens
  aggregation: sum
  dimensions:
  - account
  - model
  - api
- name: dedicated_gpu_hours
  description: GPU hours consumed by dedicated inference / compute deployments, billed per minute by hardware instance type.
  unit: gpu_hours
  aggregation: sum
  dimensions:
  - account
  - deployment
  - hardware_instance
- name: credits
  description: CentML credits consumed, where 1 credit equals 1 USD.
  unit: credits
  aggregation: sum
  dimensions:
  - account
principles:
- name: Visibility
  description: Track credit balance and burn; inspect per-model serverless token usage and per-deployment GPU-hour consumption.
- name: Allocation
  description: Tag API keys and deployments per workload/team; map serverless token spend and dedicated GPU hours to cost centers.
- name: Optimization
  description: Use serverless for spiky/low-volume workloads and dedicated deployments for sustained high throughput; right-size GPU hardware and autoscaling replicas; pick smaller models when sufficient.
- name: Accountability
  description: Assign owners per deployment/key; review credit spend monthly against budget; pause idle dedicated deployments.
maintainers:
- FN: Kin Lane
  email: kin@apievangelist.com