gram-tenant

Track LLM costs per customer. Budget enforcement, usage dashboards, billing exports.

import OpenAI from 'openai';
import { createTenantClient } from 'gram-tenant';

const client = createTenantClient(new OpenAI(), {
  apiKey: process.env.GRAM_API_KEY,
  tenantId: 'my-app',
  userId: currentUser.id,
});

// Use like normal — costs are tracked automatically
await client.chat.completions.create({
  model: 'gpt-4o',
  messages: [{ role: 'user', content: 'Hello' }]
});

Get Started

1. Sign up (coming soon)
2. Get your API key from the dashboard
3. Install the SDK:

npm install gram-tenant

Features

• Budget enforcement — Block requests when users exceed limits
• Usage tracking — Every request logged with cost, tokens, model
• Per-user dashboards — See who's spending what

Configuration

const client = createTenantClient(openai, {
  apiKey: process.env.GRAM_API_KEY,  // Required
  tenantId: 'my-app',                 // Required — your app identifier
  userId: currentUser.id,             // Required — the end user
  
  // Optional callbacks
  onWarning: (budget) => console.log('Approaching limit:', budget),
  onBlocked: (budget) => console.log('Budget exceeded:', budget),
  onError: (err) => console.error('API error:', err),
});

Error Handling

import { GramBudgetExceededError } from 'gram-tenant';

try {
  await client.chat.completions.create({...});
} catch (error) {
  if (error instanceof GramBudgetExceededError) {
    // User exceeded their budget
    console.log(error.budget); // { limit, spent, remaining, percent }
  }
}

How It Works

gram-tenant wraps your AI client:

1. Before request — Checks user's budget (cached, fail-open)
2. After request — Logs usage to your dashboard (async, non-blocking)

Built on gram-middleware for cost estimation across 69+ models.

License

MIT