anyai

One API. Any AI provider.

A provider-agnostic AI runtime for Node.js. Switch providers by configuration, not code changes.

Installation

npm install anyai

Install the provider SDK you need (optional):

# For Gemini
npm install @google/generative-ai

# For OpenAI
npm install openai

Note: anyai uses a tree-shakeable architecture. You only need to install the SDKs for the providers you actually use. Unused providers will not be included in your final bundle.

Quick Start

import { AI } from "anyai";

const ai = await AI.create({
  provider: "gemini",
  apiKey: process.env.GEMINI_API_KEY!,
  model: "gemini-2.5-flash", // ← fully type-safe, autocompletes Gemini models only
});

const response = await ai.chat.send({
  messages: [{ role: "user", content: "Hello!" }],
});

console.log(response.message.content);

Type-Safe Provider Models

When you set provider, the model field is constrained to only that provider's models:

// ✅ Type-safe — only Gemini models allowed
await AI.create({
  provider: "gemini",
  model: "gemini-2.5-flash",
  apiKey: "...",
});

// ✅ Type-safe — only OpenAI models allowed
await AI.create({ provider: "openai", model: "gpt-4o", apiKey: "..." });

// ❌ Type error — "gpt-4o" is not a Gemini model
await AI.create({ provider: "gemini", model: "gpt-4o", apiKey: "..." });

Streaming

const stream = ai.chat.stream({
  messages: [{ role: "user", content: "Tell me a story" }],
});

for await (const chunk of stream) {
  if (chunk.type === "delta") {
    process.stdout.write(chunk.delta);
  }
}

Supported Providers

Provider	`send()`	`stream()`	SDK
Gemini	✅	✅	`@google/generative-ai`
OpenAI	✅	✅	`openai`
Anthropic	🔜	🔜	—

Error Handling

import { AnyAIProviderError, AnyAIConfigError } from "anyai";

try {
  await ai.chat.send({ messages: [{ role: "user", content: "Hello" }] });
} catch (error) {
  if (error instanceof AnyAIProviderError) {
    console.error(`Provider ${error.provider} failed:`, error.message);
  }
}

Non-Goals

This is infrastructure, not product logic. anyai does not provide:

Utility functions or helpers
Provider shortcuts (ai.gemini())
Middleware or plugins
Multimodal, images, tools, or embeddings (v1)