Skip to main content
Margin intelligence is unique to Nozle — no other billing platform offers it. The problem: AI companies bill customers per API call, per token, or per model use. But the cost to serve varies dramatically by model (GPT-4 vs GPT-3.5), by usage pattern, and by customer. Without margin visibility, you’re flying blind. What Nozle provides:
  • Per-customer margin: how much you earn vs. spend on each customer
  • Per-model margin: which LLM models are profitable vs. loss-leaders
  • Per-metric margin: margin breakdown by billable metric
  • Per-plan margin: which pricing tiers are profitable
  • Trend analysis: margin over time with configurable granularity
  • Margin simulation: what-if analysis for pricing changes
How it works:
  1. You define cost models (what each unit costs you)
  2. Nozle tracks usage events (what you charge customers)
  3. Nozle computes margin = revenue - cost per event, per customer, per model
  4. Data is stored for fast analytical queries
  5. Query via SDK (nozle.margin.) or API (GET /api/v1/margin/)
Access: All margin endpoints require a secret key (sk_). Margin data is server-side only — it should not be exposed to customers. The can() endpoint returns margin fields (cost_per_use_cents, revenue_per_use_cents, margin_per_use_cents) so you can make real-time decisions based on profitability.