Budgets
FORG tracks AI spend in real time across multiple dimensions — per user, per team, per project, and per org. Budget enforcement is implemented through rules, but the underlying spend tracking is automatic and requires no configuration.
Spend dimensions
Every signal contributes to the following spend counters simultaneously:
- User — total spend by the developer who generated the signal
- Team — total spend by the team(s) the developer belongs to
- Project — total spend tagged to the project context
- Session — total spend within the current developer session
- Org — total spend across all users
Budget windows
Spend counters are maintained for multiple rolling and calendar windows. Rules evaluate against the window you specify. Common patterns:
| Use case | Recommended window |
|---|---|
| Monthly billing cap | month (resets on 1st) |
| Daily spike detection | 1d (rolling 24h) |
| Task-level limit | session |
| Hourly rate limit | 1h (rolling) |
Cost calculation
Cost is calculated per signal using the formula:
cost_usd = (tokens_in * price_per_input_token)
+ (tokens_out * price_per_output_token)Pricing tables are maintained by FORG and updated automatically. Current model prices are visible in Dashboard → Settings → Pricing. Custom model pricing can be configured for internal or self-hosted models.
Budget vs. billing
FORG budget tracking reflects your spend with AI providers (Anthropic, OpenAI, Google, etc.) — not your FORG subscription cost. FORG observes the tokens and models used and calculates the estimated provider cost. Actual charges appear on your provider bills, not from FORG.
Viewing budgets
Live budget status is available in Dashboard → Analytics → Budgets. For each user and team, FORG shows spend to date in the current month and a projected month-end total based on the daily run rate.
The Reports API provides programmatic access to the same data for custom dashboards and alerts.