Skip to content

alfianlosari/chatgpt_api_dart

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

15 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

ChatGPT Client

Access OpenAI ChatGPT Official API using Dart Language. Supports any Dart project and all Flutter target platforms (iOS, Android, Windows, Linux, Web)

Features

  • Using OpenAI ChatGPT Official Completions API Endpoint.
  • Maintain Chat History List on the client (~4000 tokens) so new prompt is aware of the previous chat context.
  • Support any custom model as Parameter (GPT-3.5, GPT-4, etc)
  • Support Stream HTTP Response using native Dart Stream.
  • Support Standard HTTP Response.
  • Pass custom system prompt and temperature.

Getting started

Register for API key from OpenAI.

Installation

This package is hosted in Pub Dev Library

Flutter App

flutter pub add chagpt_client

Dart App

dart pub add chagpt_client

Usage

Initialization

Initialize with api key. Default model is gpt-3.5-turbo.

import 'package:chatgpt_client/chatgpt_client.dart';

const client = ChatGPTClient(apiKey: "API_KEY");

optionally, you can provide the system prompt, temperature, and model like so.

const client = ChatGPTClient(apiKey: "API_KEY",
    model: "gpt-4",
    systemPrompt: "You are a CS Professor",
    temperature: 0.7);

There are 2 APIs: stream and normal

Stream Response

The server will yield chunks of data until the stream completes or throws an error.

try {
    var text = "";
    final stream = client.sendMessageStream(prompt);
    await for (final textChunk in stream) {
        text += textChunk;
        print(textChunk);
    }
    print(text);
} catch (exception) {
    print(exception.toString());
}

Normal

A normal HTTP request and response lifecycle. Server will send the complete text (it will take more time to response)

try {
    final text = await client.sendMessage(prompt);
    print(text);
} catch (exception) {
    print(exception.toString());
}   

History List

The client stores the history list of the conversation that will be included in the new prompt so ChatGPT aware of the previous context of conversation. When sending new prompt, the client will make sure the token is not exceeding 4000 (using calculation of 1 token=4chars), in case it exceeded the token, some of previous conversations will be truncated

You can also delete the history list by invoking

client.clearHistoryList();