py-gpt  by szczyglis-dev

Desktop AI assistant for multimodal interaction with various LLMs

created 2 years ago
1,142 stars

Top 34.4% on sourcepulse

GitHubView on GitHub
Project Summary

PyGPT is a comprehensive desktop AI assistant designed for users who want to interact with various large language models (LLMs) locally or via APIs. It offers a rich feature set for chat, image generation/analysis, voice control, and system automation, aiming to provide a powerful and flexible AI experience on Linux, Windows, and macOS.

How It Works

PyGPT integrates multiple LLM providers, including OpenAI (GPT-4o, GPT-4 Vision, GPT-3.5), Google Gemini, Anthropic Claude, and local models via Ollama (Llama 3, Mistral). It leverages LangChain and LlamaIndex for broader model compatibility and advanced features like "Chat with Files," enabling users to query documents, webpages, and other data sources. The application supports various interaction modes, from basic chat to complex agent-based workflows, and offers extensive plugin support for extending functionality.

Quick Start & Requirements

  • Installation: Binaries for Linux (zip) and Windows (msi) are available. Alternatively, install via pip install pygpt-net or run from source.
  • Requirements: Python >=3.10, <3.13. OpenAI API key required for OpenAI models; local models do not require API keys. Linux binaries require GLIBC >= 2.35.
  • Documentation: https://pygpt.readthedocs.io
  • Downloads: https://pygpt.net/#download

Highlighted Details

  • Supports 12 operational modes including Chat, Vision, Image Generation, LangChain, Chat with Files, Agents, and Autonomous Mode.
  • Extensive plugin system for voice control, code interpretation, file I/O, web search, and more.
  • Advanced memory management with context saving/loading and automatic summarization.
  • Accessibility features including voice control and customizable keyboard shortcuts.

Maintenance & Community

Licensing & Compatibility

  • MIT License.
  • Compatible with commercial use and closed-source linking.

Limitations & Caveats

  • Some features, like local model support for audio/vision loaders, are only available in the Python/PyPi version, not compiled or Snap versions.
  • Autonomous modes and agents are noted as experimental or beta.
Health Check
Last commit

11 hours ago

Responsiveness

1 day

Pull Requests (30d)
1
Issues (30d)
18
Star History
157 stars in the last 90 days

Explore Similar Projects

Starred by Chip Huyen Chip Huyen(Author of AI Engineering, Designing Machine Learning Systems), Andre Zayarni Andre Zayarni(Cofounder of Qdrant), and
2 more.

RealChar by Shaunwei

0.1%
6k
Real-time AI character/companion creation and interaction codebase
created 2 years ago
updated 1 year ago
Feedback? Help us improve.