vantageaiops

vantageaiops

LLM cost tracking and AI API monitoring SDK for TypeScript and JavaScript.

Track token usage, cost, latency and quality for OpenAI, Anthropic, Google and Mistral — with one line of code.

Install

npm install vantageaiops
# peer deps — install whichever providers you use
npm install openai              # for OpenAI proxy
npm install @anthropic-ai/sdk   # for Anthropic proxy

Quickstart — OpenAI

import { init, createOpenAIProxy } from "vantageaiops";
import OpenAI from "openai";

// 1. Init once (e.g. in app startup)
init({ apiKey: "vnt_your_key" });

// 2. Wrap your OpenAI client — zero other changes needed
const openai = createOpenAIProxy(new OpenAI());

// 3. Use normally — every call is automatically tracked
const res = await openai.chat.completions.create({
  model: "gpt-4o",
  messages: [{ role: "user", content: "Hello!" }],
});

Quickstart — Anthropic

import { init, createAnthropicProxy } from "vantageaiops";
import Anthropic from "@anthropic-ai/sdk";

init({ apiKey: "vnt_your_key" });
const client = createAnthropicProxy(new Anthropic());

const res = await client.messages.create({
  model: "claude-3-5-sonnet-20241022",
  max_tokens: 1024,
  messages: [{ role: "user", content: "Hello!" }],
});

Manual tracking

import { getClient } from "vantageaiops";

getClient().capture({
  eventId: crypto.randomUUID(),
  provider: "openai",
  model: "gpt-4o",
  promptTokens: 500,
  completionTokens: 120,
  totalCostUsd: 0.0035,
  latencyMs: 842,
  team: "search",
  environment: "production",
});

Agent / multi-step traces

import { init, trace } from "vantageaiops";
import OpenAI from "openai";

init({ apiKey: "vnt_your_key" });

const traceId = crypto.randomUUID();

// Wrap each LLM call with trace() to group them
const step1 = await trace(
  () => openai.chat.completions.create({ model: "gpt-4o", messages: [...] }),
  { traceId, spanDepth: 0, team: "agent" }
);

const step2 = await trace(
  () => openai.chat.completions.create({ model: "gpt-4o-mini", messages: [...] }),
  { traceId, spanDepth: 1, team: "agent" }
);

Traces appear in the Agent Traces tab of your dashboard with per-span cost breakdown.

Cost calculator

import { calculateCost, findCheapest } from "vantageaiops";

const cost = calculateCost("gpt-4o", 10_000, 2_000);
console.log(`Cost: $${cost.totalCostUsd.toFixed(4)}`);

const alt = findCheapest("gpt-4o", 10_000, 2_000);
console.log(`Save ${((cost.totalCostUsd - alt.costUsd) / cost.totalCostUsd * 100).toFixed(0)}% with ${alt.model}`);

Configuration

import { init } from "vantageaiops";

init({
  apiKey: "vnt_your_key",
  org: "acme",                         // auto-parsed from key if omitted
  team: "platform",                    // default team tag
  environment: "production",           // default: "production"
  ingestUrl: "https://api.vantageaiops.com", // default
  flushInterval: 2,                    // seconds between auto-flush
  batchSize: 50,                       // events per HTTP request
  debug: false,
});

Links

• Dashboard
• Full docs
• REST API reference
• Python SDK