Google TensorFlow · FinOps Profile

Google Tensorflow Finops

FOCUS-aligned FinOps for Google TensorFlow: open-source, no-cost framework. Cost surfaces are entirely the underlying compute, storage, and engineer-time consumed by self-hosted TensorFlow Serving and training pipelines, not TensorFlow license fees.

Google Tensorflow Finops is the FinOps profile for Google TensorFlow on the APIs.io network, aligned with the FinOps Foundation Framework.

It defines 5 billable meters, billed in USD, on a on-demand cycle, and pricing category open source.

The profile maps 7 FOCUS columns for cost-allocation reporting.

Tagged areas include FinOps, FOCUS, AI, Machine Learning, and Open Source.

Category: AI Infrastructure Pricing: Open Source Billing: On-Demand FOCUS v1.3
FinOpsFOCUSAIMachine LearningOpen Source

Framework Alignment

Framework
FinOps Foundation Framework
Data Spec
FOCUS v1.3

Charge Categories

Usage

FOCUS Columns

BillingCurrency
USD
ChargeCategory
Usage
InvoiceIssuerName
N/A (open source)
ProviderName
TensorFlow / Google
PublisherName
Google LLC
ServiceCategory
AI Infrastructure
ServiceName
Google TensorFlow

Meters

serving_inference_requests
Unit: request
Inference requests handled by self-hosted TensorFlow Serving (operational metric, not a billing line)
serving_compute_hours
Unit: instance-hour
Underlying compute hours consumed by TensorFlow Serving processes
training_compute_hours
Unit: instance-hour
Compute consumed during model training / fine-tuning
model_artifact_storage
Unit: GB-month
Storage occupied by trained model artifacts and checkpoints
hub_model_downloads
Unit: download
Pre-trained model downloads from TensorFlow Hub / Kaggle Models

Sources