specification: FinOps Framework
specificationVersion: '1.0'
schema: https://www.finops.org/framework/
provider: Hugging Face
providerId: hugging-face
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:
  - FinOps
  - FOCUS
  - AI
  - Machine Learning
  - Inference
description: 'FOCUS-aligned FinOps for Hugging Face: per-seat subscriptions plus pay-as-you-go inference
  credits, per-instance-hour for Inference Endpoints and Spaces, and per-TB-month storage with volume
  discounts.'
sources:
  - https://huggingface.co/pricing
  - https://huggingface.co/docs/inference-providers/pricing
  - https://huggingface.co/docs/inference-endpoints/pricing
  - https://huggingface.co/settings/billing
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: Hugging Face, Inc.
serviceCategory: AI Infrastructure
billingModel:
  pricingCategory: Subscription + Pay-As-You-Go
  billingFrequency: Monthly
  billingCurrency: USD
  chargeCategories:
    - Usage
    - Purchase
    - Credit
    - Adjustment
    - Tax
focusColumns:
  ServiceName: Hugging Face
  ServiceCategory: AI Infrastructure
  ProviderName: Hugging Face
  PublisherName: Hugging Face, Inc.
  InvoiceIssuerName: Hugging Face, Inc.
  BillingCurrency: USD
  ChargeCategory: Usage
meters:
  - name: subscription_seats
    description: PRO ($9/mo), Team ($20/seat/mo), Enterprise ($50+/seat/mo) seat fees.
    unit: seat
    aggregation: max
    dimensions:
      - plan
      - organization
  - name: inference_provider_credits
    description: Pay-as-you-go credits consumed against partner inference providers; pass-through
      provider rate, no HF markup.
    unit: USD
    aggregation: sum
    dimensions:
      - provider
      - model
      - task
      - organization
  - name: inference_endpoint_hours
    description: Replica-hours for dedicated Inference Endpoints, billed by hardware tier.
    unit: instance-hour
    aggregation: sum
    dimensions:
      - hardware
      - cloud
      - region
      - endpoint
  - name: spaces_hardware_hours
    description: Hourly cost of Spaces hardware (CPU upgrade, T4, L4, A10G, A100, etc).
    unit: instance-hour
    aggregation: sum
    dimensions:
      - hardware
      - space
      - owner
  - name: hub_storage
    description: Tiered TB-month storage for public and private repos with volume discounts at 50/200/500
      TB.
    unit: GB-month
    aggregation: avg
    dimensions:
      - repo_type
      - tier
      - owner
  - name: zerogpu_seconds
    description: H200 GPU-seconds consumed under the ZeroGPU shared pool (free tier with quotas by
      plan).
    unit: second
    aggregation: sum
    dimensions:
      - plan
      - space
principles:
  - name: Visibility
    description: Track spending in the billing settings (https://huggingface.co/settings/billing)
      and the Inference Providers usage breakdown by model and provider. Organizations see pooled
      usage on the org billing page.
  - name: Allocation
    description: Use the X-HF-Bill-To header to attribute Inference Providers requests to a specific
      organization. Tag Spaces and Inference Endpoints with metadata for team/product chargeback.
  - name: Optimization
    description: Use ":cheapest" provider routing policy for cost-sensitive workloads; bring your
      own provider key when you have negotiated rates with a specific partner; scale Inference Endpoints
      to zero when idle; use ZeroGPU for low-traffic Spaces; consolidate storage to take advantage
      of 50/200/500 TB volume discounts.
  - name: Accountability
    description: Team and Enterprise admins set spending limits and disable specific Inference Providers
      from organization settings. Reconcile monthly invoices against the per-model usage breakdown
      to attribute spend back to product owners.
maintainers:
  - FN: Kin Lane
    email: kin@apievangelist.com