Cost Metadata

Version: v1.1.0
This page documents behavior guaranteed in TealTiger v1.1.0. Every TealTiger audit event includes standardized cost metadata, making it easy to track spending across your AI applications.

Why cost metadata matters

AI costs can spiral quickly. TealTiger automatically tracks:

Estimated costs - Before making the API call
Actual costs - After the response
Budget tracking - How much budget remains
Cost attribution - Which model, user, or request spent what

This data flows into your audit logs, making cost analysis straightforward.

Standard cost fields

Every audit event includes these cost fields:

Field	Type	Description
`estimated`	number	Estimated cost before execution (USD)
`actual`	number	Actual cost after execution (USD)
`currency`	string	Currency code (default: “USD”)
`budget_scope`	string	Scope of budget (request/user/org/global)
`budget_window`	string	Time window (hourly/daily/monthly)
`budget_limit`	number	Maximum allowed spend in this window
`budget_remaining`	number	Remaining budget in this window
`risk_score`	number	Cost risk score (0-100)
`model`	string	Model identifier (e.g., “gpt-4”)
`model_tier`	string	Model tier (e.g., “premium”)

Complete example

Here’s what cost metadata looks like in a real audit event:

{
  "event": {
    "id": "evt_01HTY8K1W7Q9K9P7J9D1Y6S2ZZ",
    "type": "policy.decision",
    "timestamp": "2026-02-23T07:01:23.456Z"
  },
  "signals": {
    "cost": {
      "estimated_usd": 0.0031,
      "actual_usd": 0.0028,
      "currency": "USD",
      "tokens": {
        "input": 421,
        "output": 180,
        "total": 601
      },
      "budget": {
        "scope": "user",
        "window": "daily",
        "limit_usd": 10.0,
        "remaining_usd": 9.21,
        "used_usd": 0.79
      },
      "model": {
        "name": "gpt-4.1-mini",
        "tier": "standard",
        "provider": "openai"
      },
      "risk_score": 20
    }
  }
}

Using cost metadata

Track spending by user

// Filter audit events by user
const userCosts = auditEvents
  .filter(e => e.actor.id === 'user-123')
  .reduce((sum, e) => sum + e.signals.cost.actual_usd, 0);

console.log(`User spent: $${userCosts.toFixed(2)}`);

Monitor budget usage

// Check if user is approaching budget limit
const budget = event.signals.cost.budget;
const usagePercent = (budget.used_usd / budget.limit_usd) * 100;

if (usagePercent > 80) {
  console.warn(`User at ${usagePercent}% of daily budget`);
}

Analyze costs by model

// Group costs by model
const costsByModel = auditEvents.reduce((acc, e) => {
  const model = e.signals.cost.model.name;
  acc[model] = (acc[model] || 0) + e.signals.cost.actual_usd;
  return acc;
}, {});

console.log('Costs by model:', costsByModel);
// { "gpt-4": 12.45, "gpt-3.5-turbo": 3.21 }

Cost risk scores

TealTiger calculates a cost risk score (0-100) based on:

Budget proximity - How close to the limit
Spending velocity - Rate of spend increase
Historical patterns - Unusual spending behavior
Model tier - Premium models = higher risk

Risk Score	Level	Action
0-30	Low	Normal operation
31-60	Medium	Monitor closely
61-80	High	Alert stakeholders
81-100	Critical	Consider blocking

Best practices

Set realistic budgets - Start conservative, adjust based on data
Monitor daily - Check cost trends in your audit logs
Alert on thresholds - Set up alerts at 80% budget usage
Analyze by dimension - Break down costs by user, model, feature
Review monthly - Look for optimization opportunities

Audit Event Schema

Complete audit event structure

Logging Behavior

How TealTiger emits logs

Stop Budget Runaway

Prevent cost overruns

Decision Model

How decisions work

Getting Started

Core Concepts

Governance Domains

Policy Authoring

Architecture

Integrations

Deployment

API Reference

Cookbook

Compliance & GRC

Guides

Reference

Versions

Playground

Dashboard

Blog

Why cost metadata matters

Standard cost fields

Complete example

Using cost metadata

Track spending by user

Monitor budget usage

Analyze costs by model

Cost risk scores

Best practices

Audit Event Schema

Logging Behavior

Stop Budget Runaway

Decision Model

​Why cost metadata matters

​Standard cost fields

​Complete example

​Using cost metadata

​Track spending by user

​Monitor budget usage

​Analyze costs by model

​Cost risk scores

​Best practices

​Related reading

Audit Event Schema

Logging Behavior

Stop Budget Runaway

Decision Model

Why cost metadata matters

Standard cost fields

Complete example

Using cost metadata

Track spending by user

Monitor budget usage

Analyze costs by model

Cost risk scores

Best practices

Related reading