Control Your AI Costs: Pay-Per-Use AI Tools via API (No Surprises)

A developer recently reported spending $600 in a single session after overrunning their Cursor Max subscription by "an hour or two." Another team found their monthly AI API bill hit $2,000 before they realized their token usage had spiked.

The problem isn't AI. The problem is pricing models that charge you before you know what you'll use.

The Three AI Pricing Traps

Trap 1: Subscriptions you don't fully use You pay $99/month for Ahrefs. You ran 12 SEO analyses. That's $8.25 per analysis. With pay-per-use, those 12 analyses would cost $1.20.

Trap 2: Token-based billing you can't predict OpenAI charges per token. A "simple" article generation might cost $0.02 or $0.15 depending on length, model, and how many retries your prompt needs. You don't know until the bill arrives.

Trap 3: Overage charges with no ceiling Some platforms give you a quota, then charge premium rates for overages. You ship a feature, it goes viral, your bill goes from $50 to $5,000.

How Pay-Per-Use Fixes This

AADDYY uses a credit system. 1 credit = $0.01. You buy credits upfront. Every tool shows its cost before you call it. When credits run out, calls stop. No surprises.

import { AADDYY, InsufficientCreditsError } from 'aaddyy'
const client = new AADDYY()

try {
  const article = await client.articles.generate({ topic: 'AI trends' })
  // Cost: 5 credits ($0.05) — deducted from your balance
  // Response includes: { cost: 0.05, remainingCredits: 9.45 }
} catch (err) {
  if (err instanceof InsufficientCreditsError) {
    console.log(`Need ${err.required} credits, have ${err.available}`)
    // You know exactly what happened and why
  }
}

Every response tells you: what it cost, what's left. No end-of-month surprise.

Cost Per Feature

Budget Before You Build

You can calculate your exact costs before writing a line of code:

Monthly users: 1,000
Average actions per user: 5
Credits per action: ~5 (articles, images, etc.)
Total monthly credits: 25,000
Monthly cost: $250

That's it. No hidden token multipliers, no model surcharges, no overage fees. $250 serves 1,000 users. Scale linearly.

The Developer Experience

npm install aaddyy

import { AADDYY } from 'aaddyy'
const client = new AADDYY() // reads AADDYY_API_KEY from env

// Every call returns cost + remaining balance
const result = await client.articles.generate({ topic: 'AI costs' })
console.log(result.cost)             // 0.05
console.log(result.remainingCredits) // 49.95

No token counting. No model selection. No prompt engineering. The API handles all of that — you just pass your input and get your output.

For Agent Users

If you use Cursor, Claude Code, or Windsurf with the AADDYY MCP server, your AI agent's tool calls cost the same predictable per-call pricing. No runaway token bills.

{
  "mcpServers": {
    "aaddyy": {
      "command": "npx",
      "args": ["@aaddyy/mcp-server"],
      "env": { "AADDYY_API_KEY": "aip_your_key_here" }
    }
  }
}

Ask your agent to "generate 5 title options for my blog post" and it costs 1 credit ($0.01). Not $0.50 in GPT-4 tokens.

Getting Started

50 free credits on signup. No credit card. No subscription.

1. Sign up — 30 seconds
2. Create API key — 10 seconds
3. npm install aaddyy — 5 seconds
4. Start building — costs are transparent from the first call

• Pricing details
• API documentation
• SDK on npm