binary-husky/gpt_academic

Provides a Gradio-based web interface with hot-reloadable plugin architecture that decouples task logic from LLM backends, enabling seamless switching between multiple API providers through a unified configuration system. Core functionality spans document processing (PDF extraction, LaTeX rendering, Mermaid diagram generation) and real-time audio transcription with automatic sentence segmentation for conversational workflows. Includes specialized pipelines for academic work—arxiv paper fetching, fine-grained translation with source-target alignment, and grammar checking—integrated with internet aggregation plugins and a programmable "void terminal" for headless Python invocation of all functions.

70,217 stars.