GraphQL-AI-Mashup

GraphQL-AI-Mashup is an application that integrates various types of AI interfaces, providing a unified access method through GraphQL, and supporting features such as data streaming (stream).
New feature: Search the web in real time, by DuckDuckGo search engine.

Feature

Supports multiple AI services, such as OpenAI, Anthropic Claude, GeminiPro, etc.
Uses GraphQL to provide a unified API access, with standardized calling parameters
Supports data streaming (stream) functionality
Uses DataLoader to optimize API requests, improving performance
Can be deployed to Vercel with a single click

AI Support List

AI	Default Model	Documentation	Search web in real time
OpenAI	gpt-3.5-turbo	Openai Models	Not Yet
Cloudflare Workers AI	@cf/meta/llama-3-8b-instruct	Workers AI Models	Not Yet
Azure OpenAI	gpt-3.5-turbo	Azure Openai Models	Yes
GeminiPro	gemini-1.0-pro-latest	GeminiPro Models	Not Yet
Claude	claude-3-haiku-20240307	Claude Models	Normal: Yes Stream: No
Moonshot	moonshot-v1-8k	Moonshot Models	Normal: Yes Stream: No
Groq	mixtral-8x7b-32768	Groq Models	Not Yet
lingyiwanwu	yi-34b-chat-0205	LingYiWanWu Models	Not Yet
Baidu Ernie	ernie-3.5-4k-0205	Get API KEY & SECRET KEY Ernie Models	Not Yet
Ali Qwen	qwen-turbo	Get API KEY Subscribe to Services Qwen Models	Yes
Zhipu AI	glm-3-turbo	API & Token Zhipu Models	Not Yet

Getting Started

Install Dependencies

npm install

Set API Key Environment Variables

cp .env.example .env

Modify the API keys in the .env file.

Run Development Environment

npm run start

Local Access

http://localhost:4000/

Docker

This project can be deployed via docker, and it comes with its own Dockerfile.

Set API Key Environment Variables

cp .env.example .env

Modify the API keys in the .env file.

Docker Image Build

docker build -t graphql-ai-mashup .

Docker Run

docker run -dp 4000:4000 graphql-ai-mashup

Local Access

http://localhost:4000/

Deployment

Vercel

Due to Vercel's limitations on the edge runtime, please refer to the project AI-Mashup-Web for deployment.

Cloudflare

This project can be deployed to Cloudflare worker also. Please follow the steps.

Install and configure wrangler

npm install -g wrangler

wrangler login

clone this repository

git clone git@github.com:VeryInt/GraphQL-AI-Mashup.git

Build and deploy your Cloudflare Worker

cd GraphQL-AI-Mashup

wrangler build

wrangler deploy

Usage Examples

The following is an example of using GraphQL to access the GeminiPro interface:

Query

query MyQuery($params: ChatArgs) {
  chat(params: $params) {
    GeminiPro{text}
  }
}

Variables

{
 "params": {
  "messages": [{
    "role": "user",
    "content": "Please tell an 800-word fairy tale"
  }]
 }
}

Stream Query

query MyQuery($params: ChatArgs) {
  chat(params: $params) {
    GeminiProStream @stream
  }
}

ScreenShots