What is a Product?
A product in Lava defines the pricing and billing rules for AI services. Merchants can configure:
- Fee models: Fixed per-unit costs or percentage markups
- Tiered pricing: Volume-based discounts (e.g., first 1M tokens at X,next1MatY)
- Billing basis: Charge for input+output tokens, output-only, characters, or time
- Cost attribution: Who pays base costs and service charges (wallet or merchant)
- Overdraft behavior: Block requests or allow overdraft when funds run low
Each product has a unique product secret that can be included in forward tokens to apply specific pricing rules.
Fee Structure Options
Fixed Fee Model
Charge a flat amount per billing unit:
fee = fixedAmount × units
- $0.50 per API call (regardless of usage)
- $5 per 1M tokens (flat rate)
- $0.10 per minute of audio
Best for:
- Predictable pricing for users
- Simplified billing calculations
- Services with consistent resource usage
Percentage Markup Model
Add a percentage on top of the base provider cost:
fee = baseCost × percentageMarkup
- 20% markup on provider costs
- 50% markup for premium features
- 10% markup for resale
Best for:
- Scaling fees with actual costs
- Maintaining margins across providers
- Variable usage patterns
Hybrid Model
Combine fixed and percentage fees:
fee = fixedAmount + (baseCost × percentageMarkup)
- $0.10 base + 15% markup
- Covers processing overhead (fixed) + cost scaling (percentage)
Tiered Pricing
Create volume discounts by defining pricing tiers:
{
"tiers": [
{
"upTo": 1000000,
"fixedFee": 0.05,
"percentageFee": 0
},
{
"upTo": 10000000,
"fixedFee": 0.03,
"percentageFee": 0
},
{
"upTo": null,
"fixedFee": 0.01,
"percentageFee": 0
}
]
}
- First 1M units: $0.05 per unit
- Next 9M units (1M-10M): $0.03 per unit
- Above 10M units: $0.01 per unit
Benefits:
- Incentivize higher usage
- Reward loyal customers
- Competitive with direct provider pricing at scale
Billing Basis
Define what metric drives pricing:
| Billing Basis | Unit | Use Cases |
|---|
| input-output | Tokens | LLMs (charge for both prompt and completion) |
| output-only | Tokens | LLMs (charge only for completion) |
| characters | Characters | Text-to-speech, text processing |
| duration | Seconds/Minutes | Voice calls, real-time services |
| requests | API calls | Per-request pricing regardless of usage |
// User sends 500 token prompt, receives 200 token response
// Billing basis: output-only
// Charge: 200 tokens × rate (not 700 total)
Cost Attribution
Control who pays for different cost components:
Base Costs
Who pays the AI provider’s charges?
- Wallet pays (default): User covers provider costs
- Merchant pays: Merchant absorbs provider costs
Merchant Fees
Who pays the merchant’s markup?
- Wallet pays (always): User covers merchant fees
- Cannot be attributed to merchant (that would be self-payment)
Service Charge
Who pays Lava’s 1.9% platform fee?
- Follows merchant fees: If wallet pays fees, wallet pays service charge
- Attribution: Same as merchant fee payer
Example configurations:
| Model | Base Costs | Merchant Fees | User Pays | Merchant Pays |
|---|
| Pass-through | Wallet | Wallet | Base + Fees + Service | $0 |
| Merchant Absorbs Base | Merchant | Wallet | Fees + Service | Base |
| Freemium | Merchant | Merchant | $0 | Base + Fees + Service |
Overdraft Behavior
Configure what happens when wallet balance is insufficient:
Block Requests (Default)
- Request is rejected with insufficient funds error
- User must add funds before continuing
- Protects users from unexpected charges
Allow Overdraft
- Request proceeds even with insufficient balance
- Transfer created as “under-settled”
- Settled when user adds funds (oldest first)
- Useful for preventing service interruption
Configuration:
{
"overdraftAllowed": false,
"minimumBalance": 5.00
}
Product Secret Usage
Each product has a unique secret that can be optionally included in forward tokens:
// Without product secret (uses merchant default pricing)
const token = btoa(`${secretKey}.${connectionSecret}`);
// With product secret (applies specific product pricing)
const token = btoa(`${secretKey}.${connectionSecret}.${productSecret}`);
- A/B testing: Different pricing for user cohorts
- Premium tiers: Charge different rates based on subscription level
- Custom contracts: Special pricing for enterprise customers
- Feature gating: Different products for different features
Product secrets can be included in client-side code - they don’t grant access alone, just specify which pricing to apply.
Next Steps
