openai-fetch

OpenAI Fetch Client: A Lightweight Alternative

The OpenAI Fetch Client is a minimalist and highly efficient client for interacting with OpenAI's services. Engineered to leverage the native fetch function, it stands as a streamlined alternative to the official OpenAI package, especially when users face challenges with the latter's performance and size.

Why Choose openai-fetch?

There are several compelling reasons to opt for the openai-fetch package over the official OpenAI client:

• Efficiency and Speed: Designed to be lean and fast, openai-fetch does not modify the fetch function, ensuring compatibility with environments that support native fetch.
• Broad Compatibility: It operates seamlessly across various platforms, including Node 18+, browsers, Deno, and Cloudflare Workers.
• Smaller Footprint: With a package size of approximately 14kb, it is considerably smaller than the official OpenAI package, which is 152kb.
• Focused Functionality: It covers essential OpenAI tasks like chat, completions, embeddings, moderations, and Text-to-Speech (TTS).

When to Stick with the Official OpenAI Package

However, the official OpenAI package might still be preferable if:

• Your runtime environment lacks native fetch support.
• Your application cannot work with native ECMAScript Modules (ESM).
• You require access to additional OpenAI endpoints beyond those supported by openai-fetch.
• The larger library size or fetch method modifications are not a concern.

Installation and Setup

To integrate the OpenAI Fetch Client into your project, you simply execute:

npm install openai-fetch

This installation requires Node 18 or an environment compatible with the fetch method. The package is built using ESM, so projects using CommonJS may need to transition to ESM or utilize the dynamic import() function.

Quick Start Guide

Using the OpenAI Fetch Client is straightforward. Here's a basic example:

import { OpenAIClient } from 'openai-fetch';

const client = new OpenAIClient({ apiKey: '<your api key>' });

The API key is optional if it's already set in your environment as process.env.OPENAI_API_KEY.

Core Functionalities

The OpenAI Fetch Client mirrors OpenAI's API closely. This means most of the OpenAI reference documentation applies, providing a familiar interface for developers accustomed to OpenAI's offerings. The client supports strongly typed TypeScript, ensuring any discrepancies in implementation are immediately apparent through error checking.

Some of its key capabilities include:

• Chat Completions: Generate or stream chat completions with simple functions.
• Completions: Produce multiple text completions.
• Embeddings: Create embeddings for data analysis and machine learning.
• Content Moderation: Check for content that may violate OpenAI's guidelines.
• Text-to-Speech: Synthesize speech from text using OpenAI's TTS capabilities.

Type Definitions

The client provides TypeScript definitions, enhancing the development experience by ensuring type safety. These definitions are accessible through TSServer and the project's source files.

Conclusion

The OpenAI Fetch Client, maintained under the MIT license, is a versatile and efficient tool for developers needing a compact yet powerful way to interact with OpenAI services. By focusing on core functionalities and leveraging native fetch, it ensures quick integration and execution across multiple environments. For more details and updates, the project page is hosted on Dexa.ai.