README.md

Firebase Genkit <> OpenAI Plugin

OpenAI Community Plugin for Google Firebase Genkit

genkitx-openai is a community plugin for using OpenAI APIs with Firebase Genkit. Built by The Fire Company. 🔥

This Genkit plugin allows to use OpenAI models through their official APIs.

Supported models

The plugin supports several OpenAI models:

• GPT-4o, GPT-4 with all its variants (Turbo, Vision), and GPT-3.5 Turbo for text generation;
• DALL-E 3 for image generation;
• Text Embedding Small, Text Embedding Large, and Ada for text embedding generation;
• Whisper for speech recognition;
• Text-to-speech 1 and Text-to-speech 1 HD for speech synthesis.

Installation

Install the plugin in your project with your favorite package manager:

• npm install genkitx-openai
• yarn add genkitx-openai
• pnpm add genkitx-openai

Usage

Initialize

import dotenv from 'dotenv';
import { genkit } from 'genkit';
import openAI, { gpt35Turbo } from 'genkitx-openai';

dotenv.config();

const ai = genkit({
  plugins: [openAI({ apiKey: process.env.OPENAI_API_KEY })],
  // specify a default model if not provided in generate params:
  model: gpt35Turbo,
});

Basic examples

The simplest way to generate text is by using the generate method:

const response = await ai.generate({
  model: gpt4o
  prompt: 'Tell me a joke.',
});

console.log(response.text);

Multi-modal prompt

const response = await ai.generate({
  model: gpt4o,
  prompt: [
    { text: 'What animal is in the photo?' },
    { media: { url: imageUrl } },
  ],
  config: {
    // control of the level of visual detail when processing image embeddings
    // Low detail level also decreases the token usage
    visualDetailLevel: 'low',
  },
});
console.log(response.text);

Text Embeddings

import { textEmbeddingAda002 } from 'genkitx-openai';

const embedding = await ai.embed({
  embedder: textEmbeddingAda002,
  content: 'Hello world',
});

console.log(embedding);

Within a flow

import { z } from 'genkit';

export const jokeFlow = ai.defineFlow(
  {
    name: 'jokeFlow',
    inputSchema: z.string(),
    outputSchema: z.string(),
  },
  async (subject) => {
    const llmResponse = await ai.generate({
      prompt: `tell me a joke about ${subject}`,
    });
    return llmResponse.text;
  }
);

Tool use

import { z } from 'genkit';

// ...initialize genkit (as shown above)

const createReminder = ai.defineTool(
  {
    name: 'createReminder',
    description: 'Use this to create reminders for things in the future',
    inputSchema: z.object({
      time: z
        .string()
        .describe('ISO timestamp string, e.g. 2024-04-03T12:23:00Z'),
      reminder: z.string().describe('the content of the reminder'),
    }),
    outputSchema: z.number().describe('the ID of the created reminder'),
  },
  (reminder) => Promise.resolve(3)
);

const result = await ai.generate({
  tools: [createReminder],
  prompt: `
  You are a reminder assistant.
  If you create a reminder, describe in text the reminder you created as a response.

  Query: I have a meeting with Anna at 3 for dinner - can you set a reminder for the time?
  `,
});

console.log(result.text);

For more detailed examples and the explanation of other functionalities, refer to the examples in the official Github repo of the plugin or in the official Genkit documentation.

Contributing

Want to contribute to the project? That's awesome! Head over to our Contribution Guidelines.

Need support?

Note

This repository depends on Google's Firebase Genkit. For issues and questions related to Genkit, please refer to instructions available in Genkit's repository.

Reach out by opening a discussion on Github Discussions.

Credits

This plugin is proudly maintained by the team at The Fire Company. 🔥

License

This project is licensed under the Apache 2.0 License.