Instead of billing per seat (subscription) or per call (metered), credit-based billing sells a bundle of credits upfront. Consumers spend credits as they use features, and buy more when they run low.
On Runcept, 1 credit = $0.01. Each agent has a fixed credits_per_run price. If an agent costs 5 credits, every call deducts 5 credits from your balance.
Predictable cost for consumers: You know exactly what each run costs before you call it. No surprise bills from token overages or rate multipliers.
Flat pricing for builders: Set your price once. No need to meter tokens, compute time, or request complexity. If you charge 10 credits, you get 7 credits every run.
Buy once, use anytime: Credits don't expire. Users buy a batch when they sign up and top up when needed. No recurring billing friction.
Aligns incentives: Consumers are incentivised to use agents efficiently (each run costs credits). Builders are incentivised to deliver real value (credits only charge on success).
Consumer buys 1,000 credits ($10)
↓
Calls an agent with credits_per_run = 5
↓
On complete: 5 credits deducted from consumer balance
4 credits → builder (80%)
1 credit → Runcept (20%)
↓
On failed: 0 credits deducted (automatic refund)
|------|---------|-------|
You can top up any time. No subscription required for basic usage.
Start by calculating your LLM cost per run. A typical Claude Haiku call for a 2,000-token task costs roughly $0.001-0.003. Set credits_per_run at 3-5x your cost to give yourself margin: