Skip to main content
Muster tracks AI spend at the job level and aggregates it by agent, department, and model.

​
How costs are tracked

Send token usage with each quality signal: 
quality.emit(
    job_id=job_id,
    token_input=1247,
    token_output=312,
    model="gpt-4o",
    provider="openai",
    checks=[...]
)
Muster looks up the current per-token pricing for the model/provider combination and calculates the cost in USD. Costs are converted to your display currency using daily exchange rates.

​
Cost dashboard

The Cost Dashboard shows:
  • 30-day total spend across all agents
  • Daily trend chart — spot spending spikes
  • Department breakdown — compare Finance vs Legal vs Operations
  • Top agents by spend — identify your biggest cost drivers
  • Unattributed spend — jobs with no agent_id

​
Duplicate agents

When two agents do the same job, Muster’s similarity engine detects them and flags them as a duplicate pair. The combined cost of both agents is shown — merging them is the recommended optimisation.

​
Model pricing

Pricing data is stored in the database and updated regularly. You can also manually update pricing via Settings → Model Pricing or the API if you have negotiated rates.

​
Currency support

Muster supports display in any currency. Set your display currency in Settings → Instance. Exchange rates are fetched daily from BNM (Malaysia), Bank Indonesia, and ECB.