Skip to main content
Cost Breakdown lets you slice your AI spending by any dimension: time, provider, model, feature, environment, user, or custom context.

How to Use It

  1. Go to Cost Explorer → Breakdown
  2. Choose what to group by (provider, model, feature, etc.)
  3. Apply filters to narrow the data
  4. View the chart and table

Grouping Options

Group costs by:
DimensionQuestion It Answers
TimeHow is spending changing over time?
ProviderWhich providers cost the most?
ModelWhich models are most expensive?
FeatureWhich product features drive costs?
EnvironmentHow does production compare to staging?
UserWho are the heaviest consumers?

Multi-Level Grouping

Combine dimensions for deeper analysis:
  • Time + Provider: How does each provider’s cost change over time?
  • Feature + Model: Which models power which features?
  • Environment + Feature: Feature costs across environments

Filters

Narrow results with filters:
FilterOptions
Date RangePreset (7d, 30d, 90d) or custom
ProviderOpenAI, Anthropic, Google, etc.
ModelSpecific models
EnvironmentProduction, staging, development
FeatureYour custom feature names
ContextAny custom fields you track
Combine filters to answer specific questions. “GPT-4 costs for the chat feature in production last month” is straightforward.

Visualizations

Charts

  • Bar Chart: Compare costs across categories
  • Line Chart: Show trends over time
  • Pie/Donut: Show proportional distribution
  • Stacked: See composition and totals

Table View

Below the chart, a detailed table shows:
  • Category name
  • Total cost
  • Percentage of total
  • Transaction count
  • Trend indicator (up/down)
Click column headers to sort. Click rows to drill down.

Export

Download your analysis as CSV for spreadsheet analysis and further processing.

Common Analyses

Which feature costs the most?

  1. Group by Feature
  2. Set date range to last 30 days
  3. View the breakdown chart

Is GPT-4 worth the cost vs. GPT-3.5?

  1. Filter to the models you want to compare
  2. Group by Model
  3. Compare costs alongside quality (you’ll need to assess quality separately)
  1. Filter to Environment = production
  2. Group by Time (daily or weekly)
  3. Look for upward or downward trends

What’s driving a cost spike?

  1. Set date range around the spike
  2. Group by Feature first to see which feature spiked
  3. Then group by Model to see if it’s a model change
  4. Drill into Activity for transaction-level detail