@gameonop/infera

Infera

A lightweight, zero-dependency TypeScript/Node.js package that unifies multiple LLM providers under a single, standardized interface.

Website · npm

Install

npm install @gameonop/infera

Quick Start

import { openai, setKey } from '@gameonop/infera';

setKey('openai', 'sk-your-key');

const response = await openai.call({
  model: 'gpt-4o-mini',
  messages: [{ role: 'user', content: 'Hello!' }],
});

console.log(response.text);

Features

Feature	Description
6 Providers	OpenAI, Groq, NVIDIA NIM, Cerebras, Gemini, OpenRouter
Token Counting	Automatic fallback estimation when providers don't return usage
Auto Retry	Exponential backoff with multi-provider fallback chains
Chat Sessions	Stateful conversation history with context window guard
Cost Estimator	Per-request cost estimates for all providers
Batching	Parallel execution with concurrency control
Templates	Type-safe prompt templates with {{variable}} interpolation
Output Parser	Extract JSON from any LLM response format
SubAgents	Scoped task agents with parallel pool support
Health Checks	Verify provider connectivity and API key validity
Streaming	SSE streaming for all supported providers
Zero Dependencies	Pure Node.js — no axios, no SDKs

Supported Providers

Provider	Key	Default Model
OpenAI	OPENAI_API_KEY	gpt-4o-mini
Groq	GROQ_API_KEY	llama-3.3-70b-versatile
NVIDIA NIM	NVIDIA_API_KEY	meta/llama-3.1-8b-instruct
Cerebras	CEREBRAS_API_KEY	llama3.1-8b
Google Gemini	GEMINI_API_KEY	gemini-2.5-flash
OpenRouter	OPENROUTER_API_KEY	openai/gpt-4o-mini

Documentation

For detailed documentation, examples, and the full API reference, visit:

infera.netcanvas.live

License

Apache License 2.0 — see LICENSE for details.