perplexity_dart 1.0.5

A lightweight Dart SDK to interact with Perplexity.ai's chat completion API (streaming + model switching supported).

perplexity_dart

Perplexity Dart SDK is a lightweight and type-safe Dart client for interacting with Perplexity.ai's chat/completions API.
It supports both streaming and non-streaming responses, flexible model switching (e.g., sonar, sonar-pro, etc.), and is designed to work with Flutter apps.

This package also support perplexity_flutter

---

Features

• Streamed and full chat completion support
• Switch between models with known context lengths
• Chat roles: system, user, assistant, tool
• Image input: send images as base64 or URL directly alongside text

---

Getting Started

Add the SDK to your project:

dependencies:
  perplexity_dart: 

Direct API Usage

For more control, you can use the PerplexityClient directly:

import 'package:perplexity_dart/perplexity_dart.dart';

void main() async {
  final client = PerplexityClient(
    apiKey: 'your-api-key',
  );
  
  // Create messages
  final messages = [
    StandardMessageModel(
      role: MessageRole.system,
      content: 'Be precise and concise.',
    ),
    StandardMessageModel(
      role: MessageRole.user,
      content: 'Hello, how are you?',
    ),
  ];
  
  // Non-streaming response
  final requestModel = ChatRequestModel(
    model: PerplexityModel.sonarPro,
    messages: messages,
    stream: false,
  );
  
  final response = await client.sendMessage(requestModel: requestModel);
  print(response.content);
  
  // Streaming response
  final streamRequestModel = ChatRequestModel(
    model: PerplexityModel.sonar,
    messages: messages,
    stream: true,
  );
  
  final stream = client.streamChat(requestModel: streamRequestModel);
  
  await for (final chunk in stream) {
    print(chunk);
  }

  import 'dart:convert';
  import 'dart:io';

  // Send Image from Local  in base64.
  final bytes = File('/path/to/photo.png').readAsBytesSync();
  final base64Image = base64Encode(bytes);
  final dataUri = 'data:image/png;base64,$base64Image';

  final req = ChatRequestModel.defaultImageRequest(
    url: dataUri,
    systemPrompt: 'Describe this image.',
    imagePrompt: 'What’s happening here?',
  );


  // Send Image as Image URL
  final request = ChatRequestModel.defaultImageRequest(
    url: ['https://example.com/photo.png, data:image/png;base64,iVBORw0KGgoAAAANSUhEUgA...'],
    systemPrompt: 'You are an expert image analyst.',
    imagePrompt: 'Describe what you see here.',
    stream: false,
    model: PerplexityModel.sonarPro,
  );

  final response = await client.sendMessage(requestModel: request);
  print(response.content);
}

Available Models

The SDK supports all current Perplexity models with their context lengths:

• PerplexityModel.sonar - 128K tokens
• PerplexityModel.sonarPro - 200K tokens
• PerplexityModel.sonarDeepResearch - 128K tokens
• PerplexityModel.sonarReasoning - 128K tokens
• PerplexityModel.sonarReasoningPro - 128K tokens

🔧 Advanced Configuration

The ChatRequestModel supports various parameters for fine-tuning your requests:

final requestModel = ChatRequestModel(
  model: PerplexityModel.sonar,
  messages: messages,
  stream: true,
  maxTokens: 1000,
  temperature: 0.7,
  topP: 0.9,
  searchDomainFilter: ['example.com'],
  returnImages: false,
  returnRelatedQuestions: true,
  searchRecencyFilter: 'day',
  topK: 3,
  presencePenalty: 0.0,
  frequencyPenalty: 0.0,
);

License

This project is licensed under the MIT License - see the LICENSE file for details.