@virajmishra1/bench-openai

@virajmishra1/bench-openai

Auto-instrument the OpenAI SDK with Bench. One line.

import OpenAI from "openai";
import { observe } from "@virajmishra1/bench-sdk";
import { wrapOpenAI } from "@virajmishra1/bench-openai";

const bench = observe({ apiKey: process.env.BENCH_KEY, agent: "my-agent" });
const client = wrapOpenAI(new OpenAI(), bench);

// Every call is now a Bench task — cost auto-calculated from usage tokens
await client.chat.completions.create({
  model: "gpt-4o",
  messages: [{ role: "user", content: "hello" }],
});

Covers chat.completions.create, responses.create, and embeddings.create. Pricing built-in for all current GPT-4o, GPT-4.1, o-series, and GPT-3.5 models — override via priceTable for new SKUs.