Introduction

Introduction

Tokenlay is an AI operations platform that gives you complete control over AI usage and costs across your entire application. Implement sophisticated business logic like user-tier routing, dynamic rate limiting, and budget enforcement — all through a drop-in proxy that requires zero code changes.

Transform basic AI calls into intelligent, business-aware operations that automatically adapt to your users, features, and spending patterns.

Get started in minutes

npm install @tokenlay/sdk openai
import { TokenlayOpenAI } from "@tokenlay/sdk";
 
const client = new TokenlayOpenAI({
  providerApiKey: process.env.OPENAI_API_KEY,
  tokenlayKey: process.env.TOKENLAY_KEY,
});
 
// Works exactly like OpenAI SDK
const response = await client.chat.completions.create({
  model: "gpt-3.5-turbo",
  messages: [{ role: "user", content: "Hello!" }],
});

Follow our quick start guide to be running in minutes

What makes Tokenlay different

Instead of just tracking what happened, Tokenlay controls what happens next.

// No model specified - Tokenlay picks the right one based on your rules
const client = new TokenlayOpenAI({
  providerApiKey: process.env.OPENAI_API_KEY,
  tokenlayKey: process.env.TOKENLAY_KEY,
  metadata: {
    userId: user.id,
    tier: user.subscription, // "free", "pro", "enterprise"
    feature: "chat_completion"
  }
});
 
const response = await client.chat.completions.create({
  messages: [{ role: "user", content: "Hello!" }],
  metadata: { userId: "user_123", tier: "free" }
});
// Free user → automatically routed to gpt-3.5-turbo
// Pro user → gets gpt-4
// Enterprise → gets gpt-4-turbo with higher limits

Create rules once in the dashboard, then every request automatically follows them:

  • Free users: GPT-3.5, 50 requests/day, timeout at limit
  • Pro users: GPT-4, 500 requests/day, warning at 80%
  • Enterprise: GPT-4 Turbo, unlimited, real-time cost alerts

Your business logic becomes automatic: Route expensive requests from free users to cheaper models. Give pro users higher rate limits. Block requests when budgets are exceeded. All without changing a single line of application code.

Handle limits gracefully: When someone hits their limit, don’t just error out. Timeout temporarily, warn before blocking, or downgrade to a cheaper model. Keep your users happy while protecting your costs.

See the business impact: Track costs per customer segment, understand your AI unit economics, and spot usage patterns that matter to your bottom line.

Bring Your Own Key (BYOK)

Tokenlay uses a bring-your-own-key model where you maintain complete control of your AI provider credentials.

  • Your keys, your data — Your API keys stay under your control. Tokenlay acts as a transparent proxy without storing credentials.
  • No lock-in — Switch away anytime since you own the underlying API keys and usage data.
  • Free to use — Tokenlay is completely free. You pay providers directly at standard rates with no markup.
  • Compatible with all major providers — Works with OpenAI, Anthropic, Google, AWS Bedrock, Azure OpenAI, OpenRouter, and more.

Get your free Tokenlay key and start in minutes — no credit card required.

Quick Start