FinOps AI Expert
Cost optimization for AI workloads - model selection, GPU sizing, commitment strategies, and multi-cloud cost management
npx skills add majiayu000/claude-skill-registry --skill finops-ai-frankxai-ai-architect --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
What it does
Provides expert guidance on FinOps for AI workloads, focusing on cost optimization across model selection, infrastructure sizing, commitment strategies, and multi-cloud cost management.
How it works
- Presents a Cost Components framework dividing costs into Inference, Infrastructure, and Development.
- Supplies a pricing comparison matrix for API costs (per 1M tokens) across providers and models.
- Includes a Python-based LLMCostCalculator to estimate per-query, daily, monthly, and annual costs for given models and token usage.
- Offers a Model Selection by Use Case matrix and a Model Cascade pattern to route queries to cheaper models based on complexity and confidence.
- Describes a GPUSizer with MEMORY and MODEL_MEMORY mappings to recommend GPU configurations, returning either a single suitable GPU or a multi-GPU fallback.
- Details Commitment Strategies with a Break-Even analysis and an example for Azure PTU.
- Covers Cost Monitoring & Alerts with tagging schemes (ai_cost_tags), Terraform-based budget alerts, and Cost Dashboard Metrics.
- Outlines Cost Optimization Techniques: Prompt Engineering for Cost (prompt simplification and batching) and a Caching Strategy (semantic cache) with a Spot/Preemptible Instances section.
When to use it
- Triggered by terms like cost optimization, FinOps, AI costs, GPU costs, or token pricing to optimize AI workload expenses across models and infrastructure.
What it can touch
- Declares tools: claude-code. All cost calculations, model routing, GPU sizing, and budgeting logic refer to code blocks and example classes/functions included in the material.
Caveats
- Licensing noted as MIT. No explicit risk or limitation statements beyond code examples and assumed model capability thresholds; the information is instruction and illustrative, not a guarantee of outcomes.
# FinOps AI Expert You are an expert in Financial Operations (FinOps) for AI workloads, specializing in cost optimization across model selection, infrastructure sizing, commitment strategies, and multi-cloud cost management. ## AI Cost Components ### Cost Breakdown Framework ``` ┌─────────────────────────────────────────────────────────────────┐ │ AI WORKLOAD COST STACK │ ├─────────────────────────────────────────────────────────────────┤ │ │ │ INFERENCE COSTS (60-80% typical) │ │ ├── Token costs (input + output) │ │ ├── GPU compute time │ │ └── API call overhead │ │ │ │ INFRASTRUCTURE COSTS (15-30%) │ │ ├── GPU/Compute instances │ │ ├── Storage (models, vectors, data) │ │ ├── Networking (egress, load balancers) │ │ └── Supporting services (DBs, queues, caches) │ │ │ │ DEVELOPMENT COSTS (5-15%) │ │ ├── Training/Fine-tuning compute │ │ ├── Experimentation │ │ └── Development environments │ │ │ └─────────────────────────────────────────────────────────────────┘ ``` ## LLM Pricing Comparison ### API Pricing (Per 1M Tokens) | Provider | Model | Input | Output | Context | |----------|-------|-------|--------|---------| | **OpenAI** | GPT-4o | $2.50 | $10.00 | 128K | | **OpenAI** | GPT-4
- AI Cost Components
- Cost Breakdown Framework
- LLM Pricing Comparison
- API Pricing (Per 1M Tokens)
- Cost Per Query Estimation
- Model Selection for Cost Optimization
- Decision Matrix
- Model Cascading Pattern
- GPU Cost Optimization
- GPU Pricing Comparison
- Right-Sizing GPU Workloads
- Commitment Strategies
- Reserved Capacity Comparison
- Break-Even Analysis
What does the FinOps AI Expert skill do?
Cost optimization for AI workloads - model selection, GPU sizing, commitment strategies, and multi-cloud cost management
How do I install it?
Run `npx skills add majiayu000/claude-skill-registry --skill finops-ai-frankxai-ai-architect --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From majiayu000/claude-skill-registry, a repository with 534 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.