A practical compatibility target

InferencePass follows the OpenAI chat completions request and response shapes for the supported v1 features. It normalizes provider text, usage, errors, SSE chunks, JSON output, and tool calls.

OpenAI JavaScript SDK

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.INFERENCEPASS_API_KEY,
  baseURL: "https://api.inferencepass.com/v1",
});

const response = await client.chat.completions.create({
  model: "auto",
  messages: [{ role: "user", content: "Hello" }],
});

Stable routing selectors

Use auto, fast, or reasoning when you want the router to choose. Use a concrete published model when you need that exact model and it is healthy.

Clear boundaries

Compatibility does not mean every OpenAI product is emulated. The v1 boundary excludes images, audio, embeddings, fine-tuning, and the Responses API.