Skip to main content

Quick start

1. Install​

npm install @amit641/llmmeter

2. Wrap your client​

import OpenAI from "openai";
import { meter } from "@amit641/llmmeter";

const openai = meter(new OpenAI(), { feature: "chat" });

await openai.chat.completions.create({
  model: "gpt-4o-mini",
  messages: [{ role: "user", content: "Hello!" }],
});

By default, calls are appended to ./.llmmeter/calls.jsonl so you don't lose data on the first run.

3. Switch to SQLite for the dashboard​

import { meter } from "@amit641/llmmeter";
import { sqliteSink } from "@amit641/llmmeter/sqlite";

const openai = meter(new OpenAI(), {
  sink: sqliteSink({ filePath: "./.llmmeter/llmmeter.db" }),
  feature: "chat",
});
npx @amit641/llmmeter-cli dashboard
# → http://localhost:3737

4. Attach context per request​

import { withContext } from "@amit641/llmmeter";

await withContext({ userId: "usr_abc", feature: "search", traceId: "trc_123" }, async () => {
  await openai.chat.completions.create({ ... });
  await openai.embeddings.create({ ... });
});

Every call inside the closure inherits the context — even calls made deep inside other libraries.

5. Set a budget​

const openai = meter(new OpenAI(), {
  sink: sqliteSink({ filePath: "./.llmmeter/llmmeter.db" }),
  maxDailySpendUsd: 50,
  onBudgetExceeded: "throw",
});

When you blow past $50 for the day, llmmeter throws BudgetExceededError instead of silently making the call. Use "warn" to log instead of throwing.