Billing & Credits

Everything about how Layers billing works: the credit system, cost calculations, margins, Stripe subscriptions, and pricing synchronization.

Powered by Hustle Together AI SDK - All model pricing data is synced from the Hustle Together AI SDK every 24 hours. The SDK aggregates real-time pricing from all AI providers to ensure accurate cost calculations.

Credit System Overview#

Layers uses a unified credit system that abstracts away the complexity of different AI provider pricing models into a single balance.

Core Concept

1 credit = $0.01 USD worth of AI usage (before margin)
With the default 60% margin, each credit costs you ~$0.016.

How Token Counting Works#

Understanding the separation between what the AI SDK provides and what Layers calculates is key to understanding your billing.

The Flow

You make API request

POST /api/v1/chat
{ "model": "claude-3-5-sonnet", "messages": [...] }

Hustle Together AI SDK processes request

Returns token counts only:

• prompt_tokens: 12
• completion_tokens: 8

Layers calculates costs (YOUR BILLING)

→Looks up model pricing from registry

→

Calculates base cost in USD

(12 × $3/1M) + (8 × $15/1M) = $0.00024

→

Applies 60% margin

$0.00024 × 1.6 = $0.000384

→

Converts to credits

$0.000384 / $0.01 = 0.0384 credits

Response includes both

→Standard usage field (SDK data)

→layers field (your cost breakdown)

Complete API Response

Every Layers API response includes both the standard OpenAI-compatible usage field and a detailed Layers cost breakdown:

{
  "id": "msg_01ABC123...",
  "model": "claude-3-5-sonnet-20241022",
  "role": "assistant",
  "content": [
    {
      "type": "text",
      "text": "The capital of France is Paris."
    }
  ],
  "stop_reason": "end_turn",

  // Standard usage field (from Hustle Together AI SDK)
  "usage": {
    "prompt_tokens": 12,        // ← SDK provides this
    "completion_tokens": 8,      // ← SDK provides this
    "total_tokens": 20
  },

  // Layers cost breakdown (calculated by Layers)
  "layers": {
    "credits_used": 0.00048,     // ← Your actual charge
    "latency_ms": 847,
    "cost_breakdown": {
      "model": "claude-3-5-sonnet-20241022",
      "input_tokens": 12,        // ← Used for calculation
      "output_tokens": 8,         // ← Used for calculation
      "base_cost_usd": 0.0003,   // ← Calculated by Layers
      "margin_percent": 60,       // ← Your margin
      "margin_cost_usd": 0.00018, // ← Calculated by Layers
      "total_cost_usd": 0.00048,  // ← Calculated by Layers
      "credits": 0.00048          // ← What you pay
    }
  }
}

Key Takeaway

The SDK only provides token counts. Layers does all the pricing math:
• Looks up model pricing (synced from Hustle Together AI)
• Calculates USD cost from tokens
• Applies your margin
• Converts to credits
• Provides transparent cost breakdown

Cost Calculation#

The Formula

credits = (base_cost_usd / $0.01) × (1 + margin_percent / 100)

With the default 60% margin:
credits = (base_cost_usd / $0.01) × 1.60

Example Calculation

Using Claude Sonnet 4.5 with 1,000 input tokens and 500 output tokens:

Component	Calculation	Value
Input Cost	1,000 tokens × $3.00/1M	$0.003
Output Cost	500 tokens × $15.00/1M	$0.0075
Base Cost	$0.003 + $0.0075	$0.0105
Credits (before margin)	$0.0105 / $0.01	1.05
Margin (60%)	1.05 × 0.60	0.63
Total Credits	1.05 + 0.63	1.68 credits

Model Pricing Reference#

Dynamic Pricing

For current pricing on all 40 models, see the Pricing Dashboard which pulls real-time data from the Hustle Together AI SDK. The examples below are for illustration.

Example prices per 1,000 tokens:

Model	Provider	Input USD	Output USD	Credits (60%)
GPT-4o Mini	OpenAI	$0.0001	$0.0006	~0.01
Gemini 2.5 Flash	Google	$0.0003	$0.0025	~0.05
Claude 4.5 Haiku	Anthropic	$0.001	$0.005	~0.10
Claude 4.5 Sonnet	Anthropic	$0.003	$0.015	~0.29
GPT-4o	OpenAI	$0.0025	$0.01	~0.20
Claude 4.5 Opus	Anthropic	$0.005	$0.025	~0.48

View live pricing at /dashboard/pricing.

Margin System#

The margin is the markup Layers applies on top of base AI provider costs. The default margin is 60%.

Margin Breakdown in API Response

Every API response includes both standard usage and a detailed Layers cost breakdown:

{
  // ... message content ...

  "usage": {
    "prompt_tokens": 1000,
    "completion_tokens": 500,
    "total_tokens": 1500
  },

  "layers": {
    "credits_used": 1.68,
    "latency_ms": 1243,
    "cost_breakdown": {
      "model": "claude-3-5-sonnet-20241022",
      "input_tokens": 1000,
      "output_tokens": 500,
      "base_cost_usd": 0.0105,
      "margin_percent": 60,
      "margin_cost_usd": 0.0063,
      "total_cost_usd": 0.0168,
      "credits": 1.68
    }
  }
}

Credit-to-Cost Quick Reference

Base Cost (USD)	Credits @ 60% margin
$0.001	0.16 credits
$0.005	0.80 credits
$0.01	1.60 credits
$0.05	8.00 credits
$0.10	16.00 credits

Subscription Tiers#

Tier	Monthly Price	Credits	Rate Limit	Overage
Free	$0	50	10/min	N/A
Starter	$20/mo	500	60/min	$0.05/credit
Pro	$100/mo	3,000	300/min	$0.04/credit
Team	$200/mo	7,500	1,000/min	$0.033/credit

What Can You Do With Credits?

Credits	Approximate Usage
50 (Free)	~15 Claude Haiku conversations
500 (Starter)	~100 GPT-4o queries or ~160 Gemini Flash
3,000 (Pro)	~600 Claude Sonnet or ~1,000 GPT-4o
7,500 (Team)	~1,500 Claude Sonnet or ~2,500 GPT-4o

Stripe Integration#

Layers uses Stripe for all subscription management and billing.

Subscription Flow

Dashboard

Checkout

Stripe

Checkout

Payment

Complete

Webhook

Handler

Update

Credits

Webhook Events

Event	Action
`checkout.session.completed`	Create subscription, link customer
`customer.subscription.created`	Grant initial credits
`customer.subscription.updated`	Update tier (upgrade/downgrade)
`customer.subscription.deleted`	Revert to free tier
`invoice.paid`	Add monthly credits

Pricing Sync#

Layers automatically syncs pricing from Hustle Together AI to ensure costs match the latest rates:

Source: https://ai.hustletogether.com/api/pricing
Frequency: Daily sync via GitHub Actions
Cache: 24-hour in-memory cache
Models: 40 total (19 language, 21 image/multimodal)
Providers: 6 (Anthropic, OpenAI, Google, Perplexity, Morph, BFL, Recraft)

For current pricing on all models, see the Pricing Dashboard which displays real-time data from the Hustle Together AI SDK.

Pricing API Endpoints

Endpoint	Method	Purpose
`/api/v1/pricing`	GET	View current pricing and sync status
`/api/v1/pricing`	POST	Force refresh from Hustle Together AI

Credit Deduction Flow#

Request Flow

Pre-Flight Estimate

estimateCredits(model, max_tokens)

→ Check user balance ≥ estimate

→ Return 402 if insufficient

AI Gateway Call

→ Call AI provider

→ Get actual token usage

Actual Calculation

calculateCreditsWithBreakdown(model, input_tokens, output_tokens)

Deduction & Logging

deductCredits(userId, credits)

logUsage(userId, model, tokens, credits)

External Cost Validation#

If you calculate your own base costs, you can pass them to Layers for validation:

{
  "model": "anthropic/claude-sonnet-4.5",
  "messages": [...],
  "mirror_factory": {
    "base_cost_usd": 0.0105
  }
}

Layers validates your cost against its own calculation and flags discrepancies>5% as warnings.

Next Steps#

Architecture

How the gateway works

API Reference

Full endpoint documentation

Pricing Dashboard

View live pricing