specification: FinOps Framework
specificationVersion: '1.0'
schema: https://www.finops.org/framework/
provider: SF Compute
providerId: sfcompute
created: '2026-06-21'
modified: '2026-06-21'
reconciled: false
tags:
- GPU
- Compute
- Marketplace
- H100
- Spot Pricing
- FinOps
- Cost Management
- FOCUS
description: >-
  FinOps view of SF Compute spend. SF Compute is a spot-priced GPU market: cost
  is the market-clearing price per block of GPU-hours rather than a fixed
  per-GPU-hour rate. Buyers place priced buy orders that resolve into contracts;
  unused contract time can be resold, creating an offsetting credit. Spend is
  driven by GPU-hours consumed across instance types (h100i, h100v, h200ki),
  the per-block clearing price, and any inference-fleet capacity. Balances and
  transactions are exposed in cents via the API for reconciliation.
notes: >-
  Per-block prices float with the market and are not reconciled here; pull
  /v0/balance and /v0/transactions for actuals and use /v0/quote for forward
  cost estimates. reconciled=false because rates are market-driven.
sources:
- https://sfcompute.com/prices
- https://docs.sfcompute.com/docs/how-the-market-works
- 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: SF Compute
serviceCategory: Compute
billingModel:
  pricingCategory: Usage-Based
  billingFrequency: Usage
  billingCurrency: USD
  chargeCategories:
  - Usage
  - Purchase
  - Credit
  - Adjustment
focusColumns:
  ServiceName: SF Compute GPU Market
  ServiceCategory: Compute
  ProviderName: SF Compute
  PublisherName: SF Compute
  InvoiceIssuerName: SF Compute
  BillingCurrency: USD
  ChargeCategory: Usage
  PricingCategory: Usage-Based
meters:
- name: gpu_hours
  description: GPU-hours consumed against filled buy orders, priced per block at the market-clearing rate.
  unit: gpu_hours
  aggregation: sum
  dimensions:
  - account
  - instance_type
  - contract
- name: block_price_cents
  description: Total clearing price in cents for a purchased block of compute.
  unit: cents
  aggregation: sum
  dimensions:
  - account
  - order
  - instance_type
- name: resale_credit_cents
  description: Credit in cents from reselling unused contract time back to the market.
  unit: cents
  aggregation: sum
  dimensions:
  - account
  - contract
- name: inference_gpu_hours
  description: GPU-hours backing the OpenAI-compatible inference fleet.
  unit: gpu_hours
  aggregation: sum
  dimensions:
  - account
  - model
- name: balance_cents
  description: Available and current account balance in cents.
  unit: cents
  aggregation: last
  dimensions:
  - account
principles:
- name: Visibility
  description: Pull /v0/balance and /v0/transactions; inspect GPU-hours and per-block clearing prices.
- name: Allocation
  description: Tag API tokens and contracts per team/workload; map clusters to internal cost centers.
- name: Optimization
  description: Buy at off-peak market prices, right-size duration and node count, and resell unused contract time to recover spend.
- name: Accountability
  description: Assign owners per contract/token; review market spend and resale credits against budget.
maintainers:
- FN: Kin Lane
  email: kin@apievangelist.com