Economics & Budgeting

Layerr does not just route to the cheapest model. It routes to the most cost-effective model for the specific workload, and tracks every penny saved in the process. The economics system is built into the runtime, not bolted on as an afterthought.

Key Files

File	Purpose
`src/features/providers/economics/ProviderEconomicsOverlay.tsx`	Provider cost overlay UI
`src/replay/ReplayEconomicOverlay.tsx`	Per-trace cost breakdown

API Endpoints

Endpoint	What it does
`GET /api/economics/trace/:traceId`	Cost breakdown for a single trace
`GET /api/economics/profiles`	Available cost profiles
`GET /api/economics/simulate/:traceId`	Simulate cost with a different strategy
`GET /api/economics/simulations`	List past simulations
`GET /api/economics/savings`	Total savings summary
`GET /api/economics/savings/providers`	Savings broken down by provider
`GET /api/economics/savings/strategies`	Savings broken down by strategy
`GET /api/economics/savings/monthly`	Monthly savings trend
`GET /api/economics/insights`	Cost intelligence and recommendations
`GET /api/budgets`	Budget configuration
`GET /api/budgets/events`	Budget threshold events
`GET /api/cost/dashboard`	Cost dashboard data
`GET /api/cost/intelligence`	Pricing intelligence feed

How Savings Are Computed

Baseline: What would the request have cost with the default provider (e.g., GPT-4o)
Actual: What the routed request actually cost
Savings: baseline - actual

If the routed request costs more than baseline (rare, but happens for quality-critical tasks), it is recorded as a premium spend with an explicit rationale.

Budget Protection

Soft limits: Warnings when budget thresholds approach
Hard limits: Request rejection when budgets are exhausted
Profile-based budgets: Different budgets per workspace profile
Provider-level budgets: Cap spend on expensive providers

Cost Intelligence

The system tracks:

Provider pricing changes: Detects when a provider changes their pricing
Model deprecation: Warns when a model is being phased out
Cheaper alternatives: Suggests cheaper models for previously expensive workloads
Trend analysis: Monthly/weekly cost trends with predictions